gstreamer/gst/gstpad.c

4574 lines
128 KiB
C
Raw Normal View History

/* GStreamer
* Copyright (C) 1999,2000 Erik Walthinsen <omega@cse.ogi.edu>
* 2000 Wim Taymans <wtay@chello.be>
*
* gstpad.c: Pads for linking elements together
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Library General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Library General Public License for more details.
*
* You should have received a copy of the GNU Library General Public
* License along with this library; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 02111-1307, USA.
*/
#include "gst_private.h"
#include "gstpad.h"
#include "gstmarshal.h"
#include "gstutils.h"
#include "gstelement.h"
#include "gstbin.h"
#include "gstscheduler.h"
#include "gstevent.h"
#include "gstinfo.h"
#include "gsterror.h"
#include "gstvalue.h"
GST_DEBUG_CATEGORY_STATIC (debug_dataflow);
#define DEBUG_DATA(obj,data,notice) G_STMT_START{\
if (!data) { \
GST_CAT_DEBUG_OBJECT (debug_dataflow, obj, "NULL data value"); \
} else if (GST_IS_EVENT (data)) { \
GST_CAT_DEBUG_OBJECT (debug_dataflow, obj, "%s event %p (type %d, refcount %d)", notice, data, \
GST_EVENT_TYPE (data), GST_DATA_REFCOUNT_VALUE (data)); \
} else { \
GST_CAT_LOG_OBJECT (debug_dataflow, obj, "%s buffer %p (size %u, refcount %d)", notice, data, \
GST_BUFFER_SIZE (data), GST_BUFFER_REFCOUNT_VALUE (data)); \
} \
}G_STMT_END
#define GST_CAT_DEFAULT GST_CAT_PADS
struct _GstPadLink
{
GType type;
gboolean bla;
gboolean srcnotify;
gboolean sinknotify;
GstPad *srcpad;
GstPad *sinkpad;
GstCaps *srccaps;
GstCaps *sinkcaps;
GstCaps *filtercaps;
GstCaps *caps;
GstPadFixateFunction app_fixate;
gboolean engaged;
GstData *temp_store; /* used only when we invented a DISCONT */
};
enum
{
TEMPL_PAD_CREATED,
/* FILL ME */
TEMPL_LAST_SIGNAL
};
static GstObject *padtemplate_parent_class = NULL;
static guint gst_pad_template_signals[TEMPL_LAST_SIGNAL] = { 0 };
GType _gst_pad_type = 0;
/***** Start with the base GstPad class *****/
static void gst_pad_class_init (GstPadClass * klass);
static void gst_pad_init (GstPad * pad);
static void gst_pad_dispose (GObject * object);
static void gst_pad_set_pad_template (GstPad * pad, GstPadTemplate * templ);
static GstCaps *_gst_pad_default_fixate_func (GstPad * pad,
const GstCaps * caps);
static gboolean gst_pad_link_try (GstPadLink * link);
static void gst_pad_link_free (GstPadLink * link);
#ifndef GST_DISABLE_LOADSAVE
static xmlNodePtr gst_pad_save_thyself (GstObject * object, xmlNodePtr parent);
#endif
static GstObject *pad_parent_class = NULL;
GType
gst_pad_get_type (void)
{
if (!_gst_pad_type) {
static const GTypeInfo pad_info = {
sizeof (GstPadClass), NULL, NULL,
(GClassInitFunc) gst_pad_class_init, NULL, NULL,
sizeof (GstPad),
0,
(GInstanceInitFunc) gst_pad_init, NULL
};
_gst_pad_type = g_type_register_static (GST_TYPE_OBJECT, "GstPad",
&pad_info, 0);
GST_DEBUG_CATEGORY_INIT (debug_dataflow, "GST_DATAFLOW",
GST_DEBUG_BOLD | GST_DEBUG_FG_GREEN, "dataflow inside pads");
}
return _gst_pad_type;
}
static void
gst_pad_class_init (GstPadClass * klass)
{
GObjectClass *gobject_class;
gobject_class = (GObjectClass *) klass;
pad_parent_class = g_type_class_ref (GST_TYPE_OBJECT);
gobject_class->dispose = GST_DEBUG_FUNCPTR (gst_pad_dispose);
}
static void
gst_pad_init (GstPad * pad)
{
/* all structs are initialized to NULL by glib */
}
static void
gst_pad_dispose (GObject * object)
{
GstPad *pad = GST_PAD (object);
gst_pad_set_pad_template (pad, NULL);
G_OBJECT_CLASS (pad_parent_class)->dispose (object);
}
/***** Then do the Real Pad *****/
/* Pad signals and args */
enum
{
REAL_LINKED,
REAL_UNLINKED,
REAL_FIXATE,
/* FILL ME */
REAL_LAST_SIGNAL
};
enum
{
REAL_ARG_0,
REAL_ARG_CAPS,
REAL_ARG_ACTIVE
/* FILL ME */
};
static void gst_real_pad_class_init (GstRealPadClass * klass);
static void gst_real_pad_init (GstRealPad * pad);
static void gst_real_pad_dispose (GObject * object);
static gboolean _gst_real_pad_fixate_accumulator (GSignalInvocationHint * ihint,
GValue * return_accu, const GValue * handler_return, gpointer dummy);
static void gst_real_pad_set_property (GObject * object, guint prop_id,
const GValue * value, GParamSpec * pspec);
static void gst_real_pad_get_property (GObject * object, guint prop_id,
GValue * value, GParamSpec * pspec);
GType _gst_real_pad_type = 0;
static GstPad *real_pad_parent_class = NULL;
static guint gst_real_pad_signals[REAL_LAST_SIGNAL] = { 0 };
GType
gst_real_pad_get_type (void)
{
if (!_gst_real_pad_type) {
static const GTypeInfo pad_info = {
sizeof (GstRealPadClass), NULL, NULL,
(GClassInitFunc) gst_real_pad_class_init, NULL, NULL,
sizeof (GstRealPad),
0,
(GInstanceInitFunc) gst_real_pad_init, NULL
};
_gst_real_pad_type = g_type_register_static (GST_TYPE_PAD, "GstRealPad",
&pad_info, 0);
}
return _gst_real_pad_type;
}
static void
gst_real_pad_class_init (GstRealPadClass * klass)
{
GObjectClass *gobject_class;
GstObjectClass *gstobject_class;
gobject_class = (GObjectClass *) klass;
gstobject_class = (GstObjectClass *) klass;
real_pad_parent_class = g_type_class_ref (GST_TYPE_PAD);
gobject_class->dispose = GST_DEBUG_FUNCPTR (gst_real_pad_dispose);
gobject_class->set_property = GST_DEBUG_FUNCPTR (gst_real_pad_set_property);
gobject_class->get_property = GST_DEBUG_FUNCPTR (gst_real_pad_get_property);
gst_real_pad_signals[REAL_LINKED] =
g_signal_new ("linked", G_TYPE_FROM_CLASS (klass), G_SIGNAL_RUN_LAST,
G_STRUCT_OFFSET (GstRealPadClass, linked), NULL, NULL,
gst_marshal_VOID__OBJECT, G_TYPE_NONE, 1, GST_TYPE_PAD);
gst_real_pad_signals[REAL_UNLINKED] =
g_signal_new ("unlinked", G_TYPE_FROM_CLASS (klass), G_SIGNAL_RUN_LAST,
G_STRUCT_OFFSET (GstRealPadClass, unlinked), NULL, NULL,
gst_marshal_VOID__OBJECT, G_TYPE_NONE, 1, GST_TYPE_PAD);
gst_real_pad_signals[REAL_FIXATE] =
g_signal_new ("fixate", G_TYPE_FROM_CLASS (klass), G_SIGNAL_RUN_LAST,
G_STRUCT_OFFSET (GstRealPadClass, appfixatefunc),
_gst_real_pad_fixate_accumulator, NULL,
gst_marshal_BOXED__BOXED, GST_TYPE_CAPS, 1,
GST_TYPE_CAPS | G_SIGNAL_TYPE_STATIC_SCOPE);
g_object_class_install_property (G_OBJECT_CLASS (klass), REAL_ARG_ACTIVE,
g_param_spec_boolean ("active", "Active", "Whether the pad is active.",
TRUE, G_PARAM_READWRITE));
g_object_class_install_property (G_OBJECT_CLASS (klass), REAL_ARG_CAPS,
g_param_spec_boxed ("caps", "Caps", "The capabilities of the pad",
GST_TYPE_CAPS, G_PARAM_READABLE));
#ifndef GST_DISABLE_LOADSAVE
gstobject_class->save_thyself = GST_DEBUG_FUNCPTR (gst_pad_save_thyself);
#endif
gstobject_class->path_string_separator = ".";
}
static gboolean
_gst_real_pad_fixate_accumulator (GSignalInvocationHint * ihint,
GValue * return_accu, const GValue * handler_return, gpointer dummy)
{
if (gst_value_get_caps (handler_return)) {
g_value_copy (handler_return, return_accu);
/* stop emission if something was returned */
return FALSE;
}
return TRUE;
}
static void
gst_real_pad_init (GstRealPad * pad)
{
pad->direction = GST_PAD_UNKNOWN;
pad->peer = NULL;
WARNING: Don't grab this updated unless you're really, REALLY sure. Original commit message from CVS: WARNING: Don't grab this updated unless you're really, REALLY sure. WARNING: Wait for the next one. Whole lotta changes here, including a few random bits: examples/*/Makefile: updated to use `libtool gcc`, not just `gcc` gst/ gstbuffer.h: updated to new flag style gst.c, gstdebug.h: added new debugging for function ptrs gstpipeline.c: set type of parent_class to the class, not the object gstthread.c: ditto plugins/ cdparanoia/cdparanoia.c: added an argument type, updated some defaults cobin/spindentity.c: updated to new do/while loopfunction style mp3encode/lame/gstlame.c: argument types, whole lotta lame options tests/: various changes Now, for the big changes: Once again, the scheduling system has changed. And once again, it broke a whole bunch of things. The gist of the change is that there is now a function pointer for gst_pad_push and gst_pad_pull, instead of a hard-wired function. Well, currently they are functions, but that's for debugging purposes only, they just call the function pointer after spewing lots of DEBUG(). This changed the GstPad structure a bit, and the GstPad API as well. Where elements used to provide chain() and pull() functions, they provide chain() and get() functions. gst_pad_set_pull[region]_function has been changed to get_pad_set_get[region]_function. This means all the elements out there that used to have pull functions need to be updated. The calls to that function have been changed in the normal elements, but the names of the functions passed is still _pull[region](), which is an aesthetic issue more than anything. As for what doesn't work yet, just about anything dealing with Connections is hosed, meaning threaded stuff won't work. This will be fixed about 12 hours from now, after I've slept, etc. The simplefake.c test works in both cothreaded and chained cases, but not much else will work due to the Connection problem. Needless to say, don't grab this unless you *need* these features *now*, else wait to update this stuff until tomorrow. I'm going to sleep now.
2000-12-16 10:18:09 +00:00
pad->chainfunc = NULL;
WARNING: Don't grab this updated unless you're really, REALLY sure. Original commit message from CVS: WARNING: Don't grab this updated unless you're really, REALLY sure. WARNING: Wait for the next one. Whole lotta changes here, including a few random bits: examples/*/Makefile: updated to use `libtool gcc`, not just `gcc` gst/ gstbuffer.h: updated to new flag style gst.c, gstdebug.h: added new debugging for function ptrs gstpipeline.c: set type of parent_class to the class, not the object gstthread.c: ditto plugins/ cdparanoia/cdparanoia.c: added an argument type, updated some defaults cobin/spindentity.c: updated to new do/while loopfunction style mp3encode/lame/gstlame.c: argument types, whole lotta lame options tests/: various changes Now, for the big changes: Once again, the scheduling system has changed. And once again, it broke a whole bunch of things. The gist of the change is that there is now a function pointer for gst_pad_push and gst_pad_pull, instead of a hard-wired function. Well, currently they are functions, but that's for debugging purposes only, they just call the function pointer after spewing lots of DEBUG(). This changed the GstPad structure a bit, and the GstPad API as well. Where elements used to provide chain() and pull() functions, they provide chain() and get() functions. gst_pad_set_pull[region]_function has been changed to get_pad_set_get[region]_function. This means all the elements out there that used to have pull functions need to be updated. The calls to that function have been changed in the normal elements, but the names of the functions passed is still _pull[region](), which is an aesthetic issue more than anything. As for what doesn't work yet, just about anything dealing with Connections is hosed, meaning threaded stuff won't work. This will be fixed about 12 hours from now, after I've slept, etc. The simplefake.c test works in both cothreaded and chained cases, but not much else will work due to the Connection problem. Needless to say, don't grab this unless you *need* these features *now*, else wait to update this stuff until tomorrow. I'm going to sleep now.
2000-12-16 10:18:09 +00:00
pad->getfunc = NULL;
pad->chainhandler = NULL;
pad->gethandler = NULL;
WARNING: Don't grab this updated unless you're really, REALLY sure. Original commit message from CVS: WARNING: Don't grab this updated unless you're really, REALLY sure. WARNING: Wait for the next one. Whole lotta changes here, including a few random bits: examples/*/Makefile: updated to use `libtool gcc`, not just `gcc` gst/ gstbuffer.h: updated to new flag style gst.c, gstdebug.h: added new debugging for function ptrs gstpipeline.c: set type of parent_class to the class, not the object gstthread.c: ditto plugins/ cdparanoia/cdparanoia.c: added an argument type, updated some defaults cobin/spindentity.c: updated to new do/while loopfunction style mp3encode/lame/gstlame.c: argument types, whole lotta lame options tests/: various changes Now, for the big changes: Once again, the scheduling system has changed. And once again, it broke a whole bunch of things. The gist of the change is that there is now a function pointer for gst_pad_push and gst_pad_pull, instead of a hard-wired function. Well, currently they are functions, but that's for debugging purposes only, they just call the function pointer after spewing lots of DEBUG(). This changed the GstPad structure a bit, and the GstPad API as well. Where elements used to provide chain() and pull() functions, they provide chain() and get() functions. gst_pad_set_pull[region]_function has been changed to get_pad_set_get[region]_function. This means all the elements out there that used to have pull functions need to be updated. The calls to that function have been changed in the normal elements, but the names of the functions passed is still _pull[region](), which is an aesthetic issue more than anything. As for what doesn't work yet, just about anything dealing with Connections is hosed, meaning threaded stuff won't work. This will be fixed about 12 hours from now, after I've slept, etc. The simplefake.c test works in both cothreaded and chained cases, but not much else will work due to the Connection problem. Needless to say, don't grab this unless you *need* these features *now*, else wait to update this stuff until tomorrow. I'm going to sleep now.
2000-12-16 10:18:09 +00:00
pad->ghostpads = NULL;
pad->caps = NULL;
pad->linkfunc = NULL;
pad->getcapsfunc = NULL;
pad->eventfunc = gst_pad_event_default;
pad->convertfunc = gst_pad_convert_default;
pad->queryfunc = gst_pad_query_default;
pad->intlinkfunc = gst_pad_get_internal_links_default;
pad->eventmaskfunc = gst_pad_get_event_masks_default;
pad->formatsfunc = gst_pad_get_formats_default;
pad->querytypefunc = gst_pad_get_query_types_default;
GST_FLAG_SET (pad, GST_PAD_DISABLED);
GST_FLAG_UNSET (pad, GST_PAD_NEGOTIATING);
gst_probe_dispatcher_init (&pad->probedisp);
}
static void
gst_real_pad_set_property (GObject * object, guint prop_id,
const GValue * value, GParamSpec * pspec)
{
g_return_if_fail (GST_IS_PAD (object));
switch (prop_id) {
case REAL_ARG_ACTIVE:
gst_pad_set_active (GST_PAD (object), g_value_get_boolean (value));
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
break;
}
}
static void
gst_real_pad_get_property (GObject * object, guint prop_id,
GValue * value, GParamSpec * pspec)
{
g_return_if_fail (GST_IS_PAD (object));
switch (prop_id) {
case REAL_ARG_ACTIVE:
g_value_set_boolean (value, !GST_FLAG_IS_SET (object, GST_PAD_DISABLED));
break;
case REAL_ARG_CAPS:
g_value_set_boxed (value, GST_PAD_CAPS (GST_REAL_PAD (object)));
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
break;
}
}
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
/* FIXME-0.9: Replace these custom functions with proper inheritance via _init
functions and object properties */
/**
* gst_pad_custom_new:
* @type: the #Gtype of the pad.
* @name: the name of the new pad.
* @direction: the #GstPadDirection of the pad.
*
* Creates a new pad with the given name and type in the given direction.
* If name is NULL, a guaranteed unique name (across all pads)
* will be assigned.
*
* Returns: a new #GstPad, or NULL in case of an error.
*/
GstPad *
gst_pad_custom_new (GType type, const gchar * name, GstPadDirection direction)
{
GstRealPad *pad;
g_return_val_if_fail (direction != GST_PAD_UNKNOWN, NULL);
pad = g_object_new (type, NULL);
gst_object_set_name (GST_OBJECT (pad), name);
GST_RPAD_DIRECTION (pad) = direction;
return GST_PAD (pad);
}
/**
* gst_pad_new:
* @name: the name of the new pad.
* @direction: the #GstPadDirection of the pad.
*
* Creates a new real pad with the given name in the given direction.
* If name is NULL, a guaranteed unique name (across all pads)
* will be assigned.
*
* Returns: a new #GstPad, or NULL in case of an error.
*/
GstPad *
gst_pad_new (const gchar * name, GstPadDirection direction)
{
return gst_pad_custom_new (gst_real_pad_get_type (), name, direction);
}
/**
* gst_pad_custom_new_from_template:
* @type: the custom #GType of the pad.
* @templ: the #GstPadTemplate to instantiate from.
* @name: the name of the new pad.
*
* Creates a new custom pad with the given name from the given template.
* If name is NULL, a guaranteed unique name (across all pads)
* will be assigned.
*
* Returns: a new #GstPad, or NULL in case of an error.
*/
GstPad *
gst_pad_custom_new_from_template (GType type, GstPadTemplate * templ,
const gchar * name)
{
GstPad *pad;
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (GST_IS_PAD_TEMPLATE (templ), NULL);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
pad = gst_pad_custom_new (type, name, templ->direction);
gst_pad_set_pad_template (pad, templ);
return pad;
}
/**
* gst_pad_new_from_template:
* @templ: the pad template to use
* @name: the name of the element
*
* Creates a new real pad with the given name from the given template.
* If name is NULL, a guaranteed unique name (across all pads)
* will be assigned.
*
* Returns: a new #GstPad, or NULL in case of an error.
*/
GstPad *
gst_pad_new_from_template (GstPadTemplate * templ, const gchar * name)
{
return gst_pad_custom_new_from_template (gst_real_pad_get_type (),
templ, name);
}
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* FIXME 0.9: GST_PAD_UNKNOWN needs to die! */
/**
* gst_pad_get_direction:
* @pad: a #GstPad to get the direction of.
*
* Gets the direction of the pad.
*
* Returns: the #GstPadDirection of the pad.
*/
GstPadDirection
gst_pad_get_direction (GstPad * pad)
{
g_return_val_if_fail (GST_IS_PAD (pad), GST_PAD_UNKNOWN);
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
if (GST_IS_REAL_PAD (pad))
return GST_PAD_DIRECTION (pad);
else
return GST_PAD_UNKNOWN;
}
/**
* gst_pad_set_active:
* @pad: the #GstPad to activate or deactivate.
* @active: TRUE to activate the pad.
*
* Activates or deactivates the given pad.
*/
void
gst_pad_set_active (GstPad * pad, gboolean active)
{
GstRealPad *realpad;
gboolean old;
GstPadLink *link;
g_return_if_fail (GST_IS_PAD (pad));
old = GST_PAD_IS_ACTIVE (pad);
if (old == active)
return;
realpad = GST_PAD_REALIZE (pad);
if (active) {
GST_CAT_DEBUG (GST_CAT_PADS, "activating pad %s:%s",
GST_DEBUG_PAD_NAME (realpad));
GST_FLAG_UNSET (realpad, GST_PAD_DISABLED);
} else {
GST_CAT_DEBUG (GST_CAT_PADS, "de-activating pad %s:%s",
GST_DEBUG_PAD_NAME (realpad));
GST_FLAG_SET (realpad, GST_PAD_DISABLED);
}
link = GST_RPAD_LINK (realpad);
if (link) {
if (link->temp_store) {
GST_CAT_INFO (GST_CAT_PADS,
"deleting cached data %p from bufpen of pad %s:%s", link->temp_store,
GST_DEBUG_PAD_NAME (realpad));
gst_data_unref (link->temp_store);
link->temp_store = NULL;
}
}
g_object_notify (G_OBJECT (realpad), "active");
}
/**
* gst_pad_set_active_recursive:
* @pad: the #GstPad to activate or deactivate.
* @active: TRUE to activate the pad.
*
* Activates or deactivates the given pad and all internally linked
* pads upstream until it finds an element with multiple source pads.
*/
void
gst_pad_set_active_recursive (GstPad * pad, gboolean active)
{
GstElement *parent;
const GList *int_links;
g_return_if_fail (GST_IS_PAD (pad));
g_return_if_fail (GST_PAD_IS_SRC (pad));
GST_CAT_DEBUG_OBJECT (GST_CAT_PADS, pad,
"Recursively %s pad %s:%s", active ? "activating" : "deactivating",
GST_DEBUG_PAD_NAME (pad));
gst_pad_set_active (pad, active);
/* If we have more than one sourcepad, then the other pads should
* possibly be kept active. FIXME: maybe we should recurse
* activation if any one pad is active and recurse deactivation
* if no single pad is active? */
parent = gst_pad_get_parent (pad);
if (!parent || parent->numsrcpads > 1)
return;
for (int_links = gst_pad_get_internal_links (pad);
int_links; int_links = g_list_next (int_links)) {
GstPad *sinkpad = GST_PAD (int_links->data);
GstPad *peer = GST_PAD_PEER (sinkpad);
GST_CAT_DEBUG_OBJECT (GST_CAT_PADS, sinkpad,
"Recursing %s on pad %s:%s",
active ? "activation" : "deactivation", GST_DEBUG_PAD_NAME (sinkpad));
gst_pad_set_active (sinkpad, active);
if (peer)
gst_pad_set_active_recursive (peer, active);
}
}
/**
* gst_pad_is_active:
* @pad: the #GstPad to query
*
* Query if a pad is active
*
* Returns: TRUE if the pad is active.
*/
gboolean
gst_pad_is_active (GstPad * pad)
{
g_return_val_if_fail (GST_IS_PAD (pad), FALSE);
return !GST_FLAG_IS_SET (pad, GST_PAD_DISABLED);
}
/**
* gst_pad_set_name:
* @pad: a #GstPad to set the name of.
* @name: the name of the pad.
*
* Sets the name of a pad. If name is NULL, then a guaranteed unique
* name will be assigned.
*/
void
gst_pad_set_name (GstPad * pad, const gchar * name)
{
g_return_if_fail (GST_IS_PAD (pad));
gst_object_set_name (GST_OBJECT (pad), name);
}
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* FIXME 0.9: This function must die */
/**
* gst_pad_get_name:
* @pad: a #GstPad to get the name of.
*
* Gets the name of a pad.
*
* Returns: the name of the pad. This is not a newly allocated pointer
* so you must not free it.
*/
const gchar *
gst_pad_get_name (GstPad * pad)
{
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
return GST_OBJECT_NAME (pad);
}
/**
WARNING: Don't grab this updated unless you're really, REALLY sure. Original commit message from CVS: WARNING: Don't grab this updated unless you're really, REALLY sure. WARNING: Wait for the next one. Whole lotta changes here, including a few random bits: examples/*/Makefile: updated to use `libtool gcc`, not just `gcc` gst/ gstbuffer.h: updated to new flag style gst.c, gstdebug.h: added new debugging for function ptrs gstpipeline.c: set type of parent_class to the class, not the object gstthread.c: ditto plugins/ cdparanoia/cdparanoia.c: added an argument type, updated some defaults cobin/spindentity.c: updated to new do/while loopfunction style mp3encode/lame/gstlame.c: argument types, whole lotta lame options tests/: various changes Now, for the big changes: Once again, the scheduling system has changed. And once again, it broke a whole bunch of things. The gist of the change is that there is now a function pointer for gst_pad_push and gst_pad_pull, instead of a hard-wired function. Well, currently they are functions, but that's for debugging purposes only, they just call the function pointer after spewing lots of DEBUG(). This changed the GstPad structure a bit, and the GstPad API as well. Where elements used to provide chain() and pull() functions, they provide chain() and get() functions. gst_pad_set_pull[region]_function has been changed to get_pad_set_get[region]_function. This means all the elements out there that used to have pull functions need to be updated. The calls to that function have been changed in the normal elements, but the names of the functions passed is still _pull[region](), which is an aesthetic issue more than anything. As for what doesn't work yet, just about anything dealing with Connections is hosed, meaning threaded stuff won't work. This will be fixed about 12 hours from now, after I've slept, etc. The simplefake.c test works in both cothreaded and chained cases, but not much else will work due to the Connection problem. Needless to say, don't grab this unless you *need* these features *now*, else wait to update this stuff until tomorrow. I'm going to sleep now.
2000-12-16 10:18:09 +00:00
* gst_pad_set_chain_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real sink #GstPad.
* @chain: the #GstPadChainFunction to set.
*
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* Sets the given chain function for the pad. The chain function is called to
* process a #GstData input buffer.
*/
void
gst_pad_set_chain_function (GstPad * pad, GstPadChainFunction chain)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
g_return_if_fail (GST_RPAD_DIRECTION (pad) == GST_PAD_SINK);
GST_RPAD_CHAINFUNC (pad) = chain;
GST_CAT_DEBUG (GST_CAT_PADS, "chainfunc for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (chain));
}
/**
* gst_pad_set_get_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real source #GstPad.
* @get: the #GstPadGetFunction to set.
*
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* Sets the given get function for the pad. The get function is called to
* produce a new #GstData to start the processing pipeline. Get functions cannot
* return %NULL.
*/
void
gst_pad_set_get_function (GstPad * pad, GstPadGetFunction get)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
g_return_if_fail (GST_RPAD_DIRECTION (pad) == GST_PAD_SRC);
GST_RPAD_GETFUNC (pad) = get;
GST_CAT_DEBUG (GST_CAT_PADS, "getfunc for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (get));
}
/**
* gst_pad_set_event_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real source #GstPad.
* @event: the #GstPadEventFunction to set.
*
* Sets the given event handler for the pad.
*/
void
gst_pad_set_event_function (GstPad * pad, GstPadEventFunction event)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
g_return_if_fail (GST_RPAD_DIRECTION (pad) == GST_PAD_SRC);
GST_RPAD_EVENTFUNC (pad) = event;
GST_CAT_DEBUG (GST_CAT_PADS, "eventfunc for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (event));
}
/**
* gst_pad_set_event_mask_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real #GstPad of either direction.
* @mask_func: the #GstPadEventMaskFunction to set.
*
* Sets the given event mask function for the pad.
*/
void
gst_pad_set_event_mask_function (GstPad * pad,
GstPadEventMaskFunction mask_func)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
GST_RPAD_EVENTMASKFUNC (pad) = mask_func;
GST_CAT_DEBUG (GST_CAT_PADS, "eventmaskfunc for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (mask_func));
}
/**
* gst_pad_get_event_masks:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a #GstPad.
*
* Gets the array of eventmasks from the given pad.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Returns: a zero-terminated array of #GstEventMask, or NULL if the pad does
* not have an event mask function.
*/
const GstEventMask *
gst_pad_get_event_masks (GstPad * pad)
{
GstRealPad *rpad;
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
rpad = GST_PAD_REALIZE (pad);
g_return_val_if_fail (rpad, NULL);
if (GST_RPAD_EVENTMASKFUNC (rpad))
return GST_RPAD_EVENTMASKFUNC (rpad) (GST_PAD (pad));
return NULL;
}
static gboolean
gst_pad_get_event_masks_dispatcher (GstPad * pad, const GstEventMask ** data)
{
*data = gst_pad_get_event_masks (pad);
return TRUE;
}
/**
* gst_pad_get_event_masks_default:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a #GstPad.
*
* Invokes the default event masks dispatcher on the pad.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Returns: a zero-terminated array of #GstEventMask, or NULL if none of the
* internally-linked pads have an event mask function.
*/
const GstEventMask *
gst_pad_get_event_masks_default (GstPad * pad)
{
GstEventMask *result = NULL;
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
gst_pad_dispatcher (pad, (GstPadDispatcherFunction)
gst_pad_get_event_masks_dispatcher, &result);
return result;
}
/**
* gst_pad_set_convert_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real #GstPad of either direction.
* @convert: the #GstPadConvertFunction to set.
*
* Sets the given convert function for the pad.
*/
void
gst_pad_set_convert_function (GstPad * pad, GstPadConvertFunction convert)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
GST_RPAD_CONVERTFUNC (pad) = convert;
GST_CAT_DEBUG (GST_CAT_PADS, "convertfunc for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (convert));
}
/**
* gst_pad_set_query_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real #GstPad of either direction.
* @query: the #GstPadQueryFunction to set.
*
* Set the given query function for the pad.
*/
void
gst_pad_set_query_function (GstPad * pad, GstPadQueryFunction query)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
WARNING: Don't grab this updated unless you're really, REALLY sure. Original commit message from CVS: WARNING: Don't grab this updated unless you're really, REALLY sure. WARNING: Wait for the next one. Whole lotta changes here, including a few random bits: examples/*/Makefile: updated to use `libtool gcc`, not just `gcc` gst/ gstbuffer.h: updated to new flag style gst.c, gstdebug.h: added new debugging for function ptrs gstpipeline.c: set type of parent_class to the class, not the object gstthread.c: ditto plugins/ cdparanoia/cdparanoia.c: added an argument type, updated some defaults cobin/spindentity.c: updated to new do/while loopfunction style mp3encode/lame/gstlame.c: argument types, whole lotta lame options tests/: various changes Now, for the big changes: Once again, the scheduling system has changed. And once again, it broke a whole bunch of things. The gist of the change is that there is now a function pointer for gst_pad_push and gst_pad_pull, instead of a hard-wired function. Well, currently they are functions, but that's for debugging purposes only, they just call the function pointer after spewing lots of DEBUG(). This changed the GstPad structure a bit, and the GstPad API as well. Where elements used to provide chain() and pull() functions, they provide chain() and get() functions. gst_pad_set_pull[region]_function has been changed to get_pad_set_get[region]_function. This means all the elements out there that used to have pull functions need to be updated. The calls to that function have been changed in the normal elements, but the names of the functions passed is still _pull[region](), which is an aesthetic issue more than anything. As for what doesn't work yet, just about anything dealing with Connections is hosed, meaning threaded stuff won't work. This will be fixed about 12 hours from now, after I've slept, etc. The simplefake.c test works in both cothreaded and chained cases, but not much else will work due to the Connection problem. Needless to say, don't grab this unless you *need* these features *now*, else wait to update this stuff until tomorrow. I'm going to sleep now.
2000-12-16 10:18:09 +00:00
GST_RPAD_QUERYFUNC (pad) = query;
GST_CAT_DEBUG (GST_CAT_PADS, "queryfunc for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (query));
}
/**
* gst_pad_set_query_type_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real #GstPad of either direction.
* @type_func: the #GstPadQueryTypeFunction to set.
*
* Set the given query type function for the pad.
*/
void
gst_pad_set_query_type_function (GstPad * pad,
GstPadQueryTypeFunction type_func)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
GST_RPAD_QUERYTYPEFUNC (pad) = type_func;
GST_CAT_DEBUG (GST_CAT_PADS, "querytypefunc for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (type_func));
}
/**
* gst_pad_get_query_types:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a #GstPad.
*
* Get an array of supported queries that can be performed
* on this pad.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Returns: a zero-terminated array of #GstQueryType.
*/
const GstQueryType *
gst_pad_get_query_types (GstPad * pad)
{
GstRealPad *rpad;
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
rpad = GST_PAD_REALIZE (pad);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (rpad, NULL);
if (GST_RPAD_QUERYTYPEFUNC (rpad))
return GST_RPAD_QUERYTYPEFUNC (rpad) (GST_PAD (pad));
return NULL;
}
static gboolean
gst_pad_get_query_types_dispatcher (GstPad * pad, const GstQueryType ** data)
{
*data = gst_pad_get_query_types (pad);
return TRUE;
}
/**
* gst_pad_get_query_types_default:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a #GstPad.
*
* Invoke the default dispatcher for the query types on
* the pad.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Returns: an zero-terminated array of #GstQueryType, or NULL if none of the
* internally-linked pads has a query types function.
*/
const GstQueryType *
gst_pad_get_query_types_default (GstPad * pad)
{
GstQueryType *result = NULL;
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
gst_pad_dispatcher (pad, (GstPadDispatcherFunction)
gst_pad_get_query_types_dispatcher, &result);
return result;
}
/**
* gst_pad_set_internal_link_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real #GstPad of either direction.
* @intlink: the #GstPadIntLinkFunction to set.
*
* Sets the given internal link function for the pad.
*/
void
gst_pad_set_internal_link_function (GstPad * pad, GstPadIntLinkFunction intlink)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
GST_RPAD_INTLINKFUNC (pad) = intlink;
GST_CAT_DEBUG (GST_CAT_PADS, "internal link for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (intlink));
}
/**
* gst_pad_set_formats_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real #GstPad of either direction.
* @formats: the #GstPadFormatsFunction to set.
*
* Sets the given formats function for the pad.
*/
void
gst_pad_set_formats_function (GstPad * pad, GstPadFormatsFunction formats)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
GST_RPAD_FORMATSFUNC (pad) = formats;
GST_CAT_DEBUG (GST_CAT_PADS, "formats function for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (formats));
}
/**
* gst_pad_set_link_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real #GstPad.
* @link: the #GstPadLinkFunction to set.
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
*
* Sets the given link function for the pad. It will be called when the pad is
* linked or relinked with caps. The caps passed to the link function are
* guaranteed to be fixed. This means that you can assume that the caps is not
* ANY or EMPTY, and that there is exactly one structure in the caps, and that
* all the fields in the structure are fixed.
*
* The return value GST_PAD_LINK_OK should be used when the caps are acceptable,
* and you've extracted all the necessary information from the caps and set the
* element's internal state appropriately.
*
* The return value GST_PAD_LINK_REFUSED should be used when the caps are
* unacceptable for whatever reason.
*
* The return value GST_PAD_LINK_DELAYED should be used when the element is in a
* state where it can't determine whether the caps are acceptable or not. This
* is often used if the element needs to open a device or process data before
* determining acceptable caps.
*
* @link must not call gst_caps_try_set_caps() on the pad that was specified as
* a parameter, although it may (and often should) call gst_caps_try_set_caps()
* on other pads.
*/
void
gst_pad_set_link_function (GstPad * pad, GstPadLinkFunction link)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
GST_RPAD_LINKFUNC (pad) = link;
GST_CAT_DEBUG (GST_CAT_PADS, "linkfunc for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (link));
}
/**
* gst_pad_set_unlink_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real #GstPad.
* @unlink: the #GstPadUnlinkFunction to set.
*
* Sets the given unlink function for the pad. It will be called
* when the pad is unlinked.
*/
void
gst_pad_set_unlink_function (GstPad * pad, GstPadUnlinkFunction unlink)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
GST_RPAD_UNLINKFUNC (pad) = unlink;
GST_CAT_DEBUG (GST_CAT_PADS, "unlinkfunc for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (unlink));
}
/**
* gst_pad_set_fixate_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real #GstPad.
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* @fixate: the #GstPadFixateFunction to set.
*
* Sets the given fixate function for the pad. Its job is to narrow down the
* possible caps for a connection. Fixate functions are called with a const
* caps, and should return a caps that is a strict subset of the given caps.
* That is, @fixate should create a caps that is "more fixed" than previously,
* but it does not have to return fixed caps. If @fixate can't provide more
* fixed caps, it should return %NULL.
*
* Note that @fixate will only be called after the "fixate" signal is emitted,
* and only if the caps are still non-fixed.
*/
void
gst_pad_set_fixate_function (GstPad * pad, GstPadFixateFunction fixate)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
GST_RPAD_FIXATEFUNC (pad) = fixate;
GST_CAT_DEBUG (GST_CAT_PADS, "fixatefunc for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (fixate));
}
/**
* gst_pad_set_getcaps_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real #GstPad.
* @getcaps: the #GstPadGetCapsFunction to set.
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
*
* Sets the given getcaps function for the pad. @getcaps should return the
* allowable caps for a pad in the context of the element's state, its link to
* other elements, and the devices or files it has opened. These caps must be a
* subset of the pad template caps. In the NULL state with no links, @getcaps
* should ideally return the same caps as the pad template. In rare
* circumstances, an object property can affect the caps returned by @getcaps,
* but this is discouraged.
*
* You do not need to call this function if @pad's allowed caps are always the
* same as the pad template caps.
*
* For most filters, the caps returned by @getcaps is directly affected by the
* allowed caps on other pads. For demuxers and decoders, the caps returned by
* the srcpad's getcaps function is directly related to the stream data. Again,
* @getcaps should return the most specific caps it reasonably can, since this
* helps with autoplugging. However, the returned caps should not depend on the
* stream type currently negotiated for @pad.
*
* Note that the return value from @getcaps is owned by the caller.
*/
void
gst_pad_set_getcaps_function (GstPad * pad, GstPadGetCapsFunction getcaps)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
GST_RPAD_GETCAPSFUNC (pad) = getcaps;
GST_CAT_DEBUG (GST_CAT_PADS, "getcapsfunc for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (getcaps));
}
/**
* gst_pad_set_bufferalloc_function:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real sink #GstPad.
* @bufalloc: the #GstPadBufferAllocFunction to set.
*
* Sets the given bufferalloc function for the pad. Note that the
* bufferalloc function can only be set on sinkpads.
*/
void
gst_pad_set_bufferalloc_function (GstPad * pad,
GstPadBufferAllocFunction bufalloc)
{
g_return_if_fail (GST_IS_REAL_PAD (pad));
g_return_if_fail (GST_PAD_IS_SINK (pad));
GST_RPAD_BUFFERALLOCFUNC (pad) = bufalloc;
GST_CAT_DEBUG (GST_CAT_PADS, "bufferallocfunc for %s:%s set to %s",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_FUNCPTR_NAME (bufalloc));
}
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* FIXME 0.9: Do we actually want to allow the case where src and sink are
switched? */
/**
* gst_pad_unlink:
* @srcpad: the source #GstPad to unlink.
* @sinkpad: the sink #GstPad to unlink.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Unlinks the source pad from the sink pad. Will emit the "unlinked" signal on
* both pads.
*/
void
gst_pad_unlink (GstPad * srcpad, GstPad * sinkpad)
{
GstRealPad *realsrc, *realsink;
GstScheduler *src_sched, *sink_sched;
g_return_if_fail (GST_IS_PAD (srcpad));
g_return_if_fail (GST_IS_PAD (sinkpad));
GST_CAT_INFO (GST_CAT_ELEMENT_PADS, "unlinking %s:%s(%p) and %s:%s(%p)",
GST_DEBUG_PAD_NAME (srcpad), srcpad,
GST_DEBUG_PAD_NAME (sinkpad), sinkpad);
realsrc = GST_PAD_REALIZE (srcpad);
realsink = GST_PAD_REALIZE (sinkpad);
g_return_if_fail (GST_RPAD_PEER (realsrc) != NULL);
g_return_if_fail (GST_RPAD_PEER (realsink) == realsrc);
if ((GST_RPAD_DIRECTION (realsrc) == GST_PAD_SINK) &&
(GST_RPAD_DIRECTION (realsink) == GST_PAD_SRC)) {
GstRealPad *temppad;
temppad = realsrc;
realsrc = realsink;
realsink = temppad;
}
g_return_if_fail ((GST_RPAD_DIRECTION (realsrc) == GST_PAD_SRC) &&
(GST_RPAD_DIRECTION (realsink) == GST_PAD_SINK));
if (GST_RPAD_UNLINKFUNC (realsrc)) {
GST_RPAD_UNLINKFUNC (realsrc) (GST_PAD (realsrc));
}
if (GST_RPAD_UNLINKFUNC (realsink)) {
GST_RPAD_UNLINKFUNC (realsink) (GST_PAD (realsink));
}
/* get the schedulers before we unlink */
src_sched = gst_pad_get_scheduler (GST_PAD (realsrc));
sink_sched = gst_pad_get_scheduler (GST_PAD (realsink));
if (GST_RPAD_LINK (realsrc))
gst_pad_link_free (GST_RPAD_LINK (realsrc));
/* first clear peers */
GST_RPAD_PEER (realsrc) = NULL;
GST_RPAD_PEER (realsink) = NULL;
GST_RPAD_LINK (realsrc) = NULL;
GST_RPAD_LINK (realsink) = NULL;
/* now tell the scheduler */
if (src_sched && src_sched == sink_sched) {
gst_scheduler_pad_unlink (src_sched, GST_PAD (realsrc), GST_PAD (realsink));
}
/* hold a reference, as they can go away in the signal handlers */
gst_object_ref (GST_OBJECT (realsrc));
gst_object_ref (GST_OBJECT (realsink));
/* fire off a signal to each of the pads telling them
* that they've been unlinked */
g_signal_emit (G_OBJECT (realsrc), gst_real_pad_signals[REAL_UNLINKED],
0, realsink);
g_signal_emit (G_OBJECT (realsink), gst_real_pad_signals[REAL_UNLINKED],
0, realsrc);
GST_CAT_INFO (GST_CAT_ELEMENT_PADS, "unlinked %s:%s and %s:%s",
GST_DEBUG_PAD_NAME (srcpad), GST_DEBUG_PAD_NAME (sinkpad));
gst_object_unref (GST_OBJECT (realsrc));
gst_object_unref (GST_OBJECT (realsink));
}
/**
* gst_pad_is_linked:
* @pad: pad to check
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Checks if a @pad is linked to another pad or not.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Returns: TRUE if the pad is linked, FALSE otherwise.
*/
gboolean
gst_pad_is_linked (GstPad * pad)
{
g_return_val_if_fail (GST_IS_PAD (pad), FALSE);
return GST_PAD_PEER (pad) != NULL;
}
static gboolean
gst_pad_check_schedulers (GstRealPad * realsrc, GstRealPad * realsink)
{
GstScheduler *src_sched, *sink_sched;
gint num_decoupled = 0;
src_sched = gst_pad_get_scheduler (GST_PAD (realsrc));
sink_sched = gst_pad_get_scheduler (GST_PAD (realsink));
if (src_sched && sink_sched) {
if (GST_FLAG_IS_SET (GST_PAD_PARENT (realsrc), GST_ELEMENT_DECOUPLED))
num_decoupled++;
if (GST_FLAG_IS_SET (GST_PAD_PARENT (realsink), GST_ELEMENT_DECOUPLED))
num_decoupled++;
if (src_sched != sink_sched && num_decoupled != 1) {
return FALSE;
}
}
return TRUE;
}
#define GST_PAD_LINK_SRC(pad) ((GST_PAD_IS_SRC (pad)) ? (pad) : GST_PAD_PEER (pad))
#define GST_PAD_LINK_SINK(pad) ((GST_PAD_IS_SINK (pad)) ? (pad) : GST_PAD_PEER (pad))
static GstPadLink *
gst_pad_link_new (void)
{
GstPadLink *link;
link = g_new0 (GstPadLink, 1);
link->sinknotify = TRUE;
link->srcnotify = TRUE;
link->engaged = FALSE;
return link;
}
static void
gst_pad_link_free (GstPadLink * link)
{
if (link->srccaps)
gst_caps_free (link->srccaps);
if (link->sinkcaps)
gst_caps_free (link->sinkcaps);
if (link->filtercaps)
gst_caps_free (link->filtercaps);
if (link->caps)
gst_caps_free (link->caps);
if (link->temp_store)
gst_data_unref (link->temp_store);
#ifdef USE_POISONING
memset (link, 0xff, sizeof (*link));
#endif
g_free (link);
}
static void
gst_pad_link_intersect (GstPadLink * link)
{
GstCaps *pad_intersection;
if (link->caps)
gst_caps_free (link->caps);
GST_DEBUG ("intersecting link from %s:%s to %s:%s",
GST_DEBUG_PAD_NAME (link->srcpad), GST_DEBUG_PAD_NAME (link->sinkpad));
GST_DEBUG ("... srccaps %" GST_PTR_FORMAT, link->srccaps);
GST_DEBUG ("... sinkcaps %" GST_PTR_FORMAT, link->sinkcaps);
GST_DEBUG ("... filtercaps %" GST_PTR_FORMAT, link->filtercaps);
pad_intersection = gst_caps_intersect (link->srccaps, link->sinkcaps);
if (link->filtercaps) {
GST_DEBUG ("unfiltered intersection %" GST_PTR_FORMAT, pad_intersection);
link->caps = gst_caps_intersect (pad_intersection, link->filtercaps);
gst_caps_free (pad_intersection);
} else {
link->caps = pad_intersection;
}
GST_DEBUG ("filtered intersection %" GST_PTR_FORMAT, link->caps);
}
static gboolean
gst_pad_link_ready_for_negotiation (GstPadLink * link)
{
GstElement *parent;
parent = GST_PAD_PARENT (link->srcpad);
if (!parent || GST_STATE (parent) < GST_STATE_READY) {
GST_DEBUG ("parent %s of pad %s:%s is not READY",
GST_ELEMENT_NAME (parent), GST_DEBUG_PAD_NAME (link->srcpad));
return FALSE;
}
parent = GST_PAD_PARENT (link->sinkpad);
if (!parent || GST_STATE (parent) < GST_STATE_READY) {
GST_DEBUG ("parent %s of pad %s:%s is not READY",
GST_ELEMENT_NAME (parent), GST_DEBUG_PAD_NAME (link->sinkpad));
return FALSE;
}
return TRUE;
}
static void
gst_pad_link_fixate (GstPadLink * link)
{
GstCaps *caps;
GstCaps *newcaps;
caps = link->caps;
g_return_if_fail (caps != NULL);
g_return_if_fail (!gst_caps_is_empty (caps));
GST_DEBUG ("trying to fixate caps %" GST_PTR_FORMAT, caps);
gst_caps_do_simplify (caps);
while (!gst_caps_is_fixed (caps)) {
int i;
for (i = 0; i < 5; i++) {
newcaps = NULL;
switch (i) {
case 0:
g_signal_emit (G_OBJECT (link->srcpad),
gst_real_pad_signals[REAL_FIXATE], 0, caps, &newcaps);
GST_DEBUG ("app srcpad signal fixated to %" GST_PTR_FORMAT, newcaps);
break;
case 1:
g_signal_emit (G_OBJECT (link->sinkpad),
gst_real_pad_signals[REAL_FIXATE], 0, caps, &newcaps);
GST_DEBUG ("app sinkpad signal fixated to %" GST_PTR_FORMAT, newcaps);
break;
case 2:
if (GST_RPAD_FIXATEFUNC (link->srcpad)) {
newcaps =
GST_RPAD_FIXATEFUNC (link->srcpad) (GST_PAD (link->srcpad),
caps);
GST_DEBUG ("srcpad %s:%s fixated to %" GST_PTR_FORMAT,
GST_DEBUG_PAD_NAME (link->srcpad), newcaps);
} else
GST_DEBUG ("srcpad %s:%s doesn't have a fixate function",
GST_DEBUG_PAD_NAME (link->srcpad));
break;
case 3:
if (GST_RPAD_FIXATEFUNC (link->sinkpad)) {
newcaps =
GST_RPAD_FIXATEFUNC (link->sinkpad) (GST_PAD (link->sinkpad),
caps);
GST_DEBUG ("sinkpad %s:%s fixated to %" GST_PTR_FORMAT,
GST_DEBUG_PAD_NAME (link->sinkpad), newcaps);
} else
GST_DEBUG ("sinkpad %s:%s doesn't have a fixate function",
GST_DEBUG_PAD_NAME (link->sinkpad));
break;
case 4:
newcaps = _gst_pad_default_fixate_func (GST_PAD (link->srcpad), caps);
GST_DEBUG ("core fixated to %" GST_PTR_FORMAT, newcaps);
break;
}
if (newcaps) {
G_GNUC_UNUSED gboolean bad;
gst_caps_do_simplify (newcaps);
#ifndef G_DISABLE_CHECKS
/* some mad checking for correctly working fixation functions */
if (i == 4) {
/* we trust the default fixation function unconditionally */
bad = FALSE;
} else if (gst_caps_is_empty (newcaps)) {
g_warning
("a fixation function did not fixate correctly, it returned empty caps");
gst_caps_free (newcaps);
continue;
} else if (gst_caps_is_any (caps)) {
bad = gst_caps_is_any (newcaps);
} else {
GstCaps *test = gst_caps_subtract (caps, newcaps);
bad = gst_caps_is_empty (test);
gst_caps_free (test);
/* simplifying is ok, too */
if (bad)
bad = (gst_caps_get_size (newcaps) >= gst_caps_get_size (caps));
}
if (bad) {
gchar *newcaps_str = gst_caps_to_string (newcaps);
gchar *caps_str = gst_caps_to_string (caps);
g_warning
("a fixation function did not fixate correctly, the returned caps %s are no true subset of %s.",
newcaps_str, caps_str);
g_free (newcaps_str);
g_free (caps_str);
gst_caps_free (newcaps);
} else
#endif
{
gst_caps_free (caps);
caps = newcaps;
break;
}
}
}
}
link->caps = caps;
}
static GstPadLinkReturn
gst_pad_link_call_link_functions (GstPadLink * link)
{
GstPadLinkReturn res = GST_PAD_LINK_OK;
/* Detect recursion. */
if (GST_PAD_IS_NEGOTIATING (link->srcpad) ||
GST_PAD_IS_NEGOTIATING (link->sinkpad)) {
GST_ERROR ("The link functions have recursed, please file a bug!");
return GST_PAD_LINK_REFUSED;
}
/* Both of the pads are in negotiation, so we set the NEGOTIATING flag on both
* of them now to avoid recursion from either pad. */
GST_FLAG_SET (link->srcpad, GST_PAD_NEGOTIATING);
GST_FLAG_SET (link->sinkpad, GST_PAD_NEGOTIATING);
/* If this doesn't run, the status is left to the default OK value. */
if (link->srcnotify && GST_RPAD_LINKFUNC (link->srcpad)) {
/* call the link function */
GST_DEBUG_OBJECT (link->srcpad,
"calling link function with caps %" GST_PTR_FORMAT, link->caps);
res = GST_RPAD_LINKFUNC (link->srcpad) (GST_PAD (link->srcpad), link->caps);
GST_DEBUG_OBJECT (link->srcpad, "got reply %d from link function", res);
if (GST_PAD_LINK_FAILED (res)) {
GST_CAT_INFO_OBJECT (GST_CAT_CAPS, link->srcpad,
"pad doesn't accept caps %" GST_PTR_FORMAT, link->caps);
}
}
if (GST_PAD_LINK_SUCCESSFUL (res) &&
link->sinknotify && GST_RPAD_LINKFUNC (link->sinkpad)) {
/* call the link function */
GST_DEBUG_OBJECT (link->sinkpad,
"calling link function with caps %" GST_PTR_FORMAT, link->caps);
res = GST_RPAD_LINKFUNC (link->sinkpad) (GST_PAD (link->sinkpad),
link->caps);
GST_DEBUG_OBJECT (link->sinkpad, "got reply %d from link function", res);
if (GST_PAD_LINK_FAILED (res)) {
GST_CAT_INFO_OBJECT (GST_CAT_CAPS, link->sinkpad,
"pad doesn't accept caps %" GST_PTR_FORMAT, link->caps);
}
}
GST_FLAG_UNSET (link->srcpad, GST_PAD_NEGOTIATING);
GST_FLAG_UNSET (link->sinkpad, GST_PAD_NEGOTIATING);
return res;
}
static GstPadLinkReturn
gst_pad_link_negotiate (GstPadLink * link)
{
GST_DEBUG ("negotiating link from pad %s:%s to pad %s:%s",
GST_DEBUG_PAD_NAME (link->srcpad), GST_DEBUG_PAD_NAME (link->sinkpad));
gst_pad_link_intersect (link);
if (gst_caps_is_empty (link->caps))
return GST_PAD_LINK_REFUSED;
gst_pad_link_fixate (link);
if (gst_caps_is_empty (link->caps))
return GST_PAD_LINK_REFUSED;
if (!gst_pad_link_ready_for_negotiation (link)) {
return GST_PAD_LINK_DELAYED;
}
GST_DEBUG ("calling link_functions between %s:%s and %s:%s with caps %"
GST_PTR_FORMAT, GST_DEBUG_PAD_NAME (link->srcpad),
GST_DEBUG_PAD_NAME (link->sinkpad), link->caps);
return gst_pad_link_call_link_functions (link);
}
/**
* gst_pad_link_try:
* @link: link to try
*
* Tries to (re)link the pads with the given link. The function takes ownership
* of the supplied link. If the function returns FALSE and an old link existed,
* that link can be assumed to work unchanged.
*
* Returns: TRUE if the link succeeded, FALSE if not.
*/
static gboolean
gst_pad_link_try (GstPadLink * link)
{
GstPad *srcpad, *sinkpad;
GstPadLink *oldlink;
GstPadLinkReturn ret;
/* we use assertions here, because this function is static */
g_assert (link);
srcpad = link->srcpad;
g_assert (srcpad);
sinkpad = link->sinkpad;
g_assert (sinkpad);
oldlink = GST_RPAD_LINK (srcpad);
g_assert (oldlink == GST_RPAD_LINK (sinkpad));
GST_DEBUG ("negotiating given link");
ret = gst_pad_link_negotiate (link);
if (GST_PAD_LINK_FAILED (ret) && oldlink && oldlink->caps) {
GST_DEBUG ("negotiating failed, but there was a valid old link");
oldlink->srcnotify = link->srcnotify;
oldlink->sinknotify = link->sinknotify;
if (GST_PAD_LINK_FAILED (gst_pad_link_call_link_functions (oldlink))) {
g_warning ("pads don't accept old caps. We assume they did though");
}
}
if (ret == GST_PAD_LINK_REFUSED) {
GST_DEBUG ("link refused, returning");
gst_pad_link_free (link);
return ret;
}
if (ret == GST_PAD_LINK_DELAYED) {
GST_DEBUG ("link delayed, replacing link caps and returning");
gst_caps_replace (&link->caps, NULL);
}
GST_RPAD_PEER (srcpad) = GST_REAL_PAD (link->sinkpad);
GST_RPAD_PEER (sinkpad) = GST_REAL_PAD (link->srcpad);
if (oldlink) {
GST_DEBUG ("copying stuff from oldlink");
link->temp_store = oldlink->temp_store;
GST_DEBUG ("moving old data temp store %p", link->temp_store);
link->engaged = oldlink->engaged;
oldlink->temp_store = NULL;
gst_pad_link_free (oldlink);
}
GST_RPAD_LINK (srcpad) = link;
GST_RPAD_LINK (sinkpad) = link;
if (ret == GST_PAD_LINK_OK) {
GST_DEBUG ("notifying caps after successful link");
g_object_notify (G_OBJECT (srcpad), "caps");
g_object_notify (G_OBJECT (sinkpad), "caps");
}
return ret;
}
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
/**
* gst_pad_renegotiate:
* @pad: a #GstPad
*
* Initiate caps negotiation on @pad. @pad must be linked.
*
* If @pad's parent is not at least in #GST_STATE_READY, returns
* #GST_PAD_LINK_DELAYED.
*
* Otherwise caps are retrieved from both @pad and its peer by calling their
* getcaps functions. They are then intersected, returning #GST_PAD_LINK_FAIL if
* there is no intersection.
*
* The intersection is fixated if necessary, and then the link functions of @pad
* and its peer are called.
*
* Returns: The return value of @pad's link function (see
* gst_pad_set_link_function()), or #GST_PAD_LINK_OK if there is no link
* function.
*
* The macros GST_PAD_LINK_SUCCESSFUL() and GST_PAD_LINK_FAILED() should be used
* when you just need success/failure information.
*/
GstPadLinkReturn
gst_pad_renegotiate (GstPad * pad)
{
GstPadLink *link;
g_return_val_if_fail (GST_IS_PAD (pad), GST_PAD_LINK_REFUSED);
if (!GST_PAD_PEER (pad))
return GST_PAD_LINK_OK;
link = gst_pad_link_new ();
link->srcpad = GST_PAD_LINK_SRC (pad);
link->sinkpad = GST_PAD_LINK_SINK (pad);
if (!gst_pad_link_ready_for_negotiation (link)) {
gst_pad_link_free (link);
return GST_PAD_LINK_DELAYED;
}
if (GST_REAL_PAD (pad)->link->filtercaps) {
link->filtercaps = gst_caps_copy (GST_REAL_PAD (pad)->link->filtercaps);
}
link->srccaps = gst_pad_get_caps (link->srcpad);
link->sinkcaps = gst_pad_get_caps (link->sinkpad);
return gst_pad_link_try (link);
}
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
/**
* gst_pad_try_set_caps:
* @pad: a #GstPad
* @caps: #GstCaps to set on @pad
*
* Try to set the caps on @pad. @caps must be fixed. If @pad is unlinked,
* returns #GST_PAD_LINK_OK without doing anything. Otherwise, start caps
* negotiation on @pad.
*
* Returns: The return value of @pad's link function (see
* gst_pad_set_link_function()), or #GST_PAD_LINK_OK if there is no link
* function.
*
* The macros GST_PAD_LINK_SUCCESSFUL() and GST_PAD_LINK_FAILED() should be used
* when you just need success/failure information.
*/
GstPadLinkReturn
gst_pad_try_set_caps (GstPad * pad, const GstCaps * caps)
{
GstPadLink *link;
GstPadLink *oldlink;
GstPadLinkReturn ret;
g_return_val_if_fail (GST_IS_REAL_PAD (pad), GST_PAD_LINK_REFUSED);
GST_LOG_OBJECT (pad, "Trying to set %" GST_PTR_FORMAT, caps);
if (GST_PAD_IS_NEGOTIATING (pad)) {
GST_DEBUG_OBJECT (pad, "Detected a recursion, just returning OK");
return GST_PAD_LINK_OK;
}
GST_CAT_INFO_OBJECT (GST_CAT_CAPS, pad, "caps %" GST_PTR_FORMAT, caps);
/* setting non-fixed caps on a pad is not allowed */
if (!gst_caps_is_fixed (caps)) {
GST_CAT_INFO (GST_CAT_CAPS,
"trying to set unfixed caps on pad %s:%s, not allowed",
GST_DEBUG_PAD_NAME (pad));
g_warning ("trying to set non fixed caps on pad %s:%s, not allowed",
GST_DEBUG_PAD_NAME (pad));
GST_DEBUG ("unfixed caps %" GST_PTR_FORMAT, caps);
return GST_PAD_LINK_REFUSED;
}
/* we allow setting caps on non-linked pads. It's ignored */
if (!GST_PAD_PEER (pad)) {
GST_DEBUG ("unlinked pad %s:%s, returning OK", GST_DEBUG_PAD_NAME (pad));
return GST_PAD_LINK_OK;
}
/* we just checked that a peer exists */
g_assert (GST_PAD_LINK_SRC (pad));
g_assert (GST_PAD_LINK_SINK (pad));
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* if the desired caps are already there, it's trivially ok */
gst/gstcaps.c: check for ANY caps before appending/unioning Original commit message from CVS: * gst/gstcaps.c: (gst_caps_append), (gst_caps_union): check for ANY caps before appending/unioning * gst/gstcaps.c: (gst_caps_is_subset), (gst_caps_is_equal), (gst_caps_structure_subtract_field), (gst_caps_structure_subtract), (gst_caps_subtract): * gst/gstcaps.h: add gst_caps_is_equal, gst_caps_is_subset and gst_caps_subtract to the API. deprecate gst_caps_is_equal_fixed * gst/gstpad.c: (gst_pad_try_set_caps): * gst/gstqueue.c: (gst_queue_link): s/gst_caps_is_equal_fixed/gst_caps_is_equal/ * gst/gststructure.c: (gst_structure_get_name_id): * gst/gststructure.h: add function gst_structure_get_name_id * gst/gstvalue.c: (gst_value_subtract_int_int_range), (gst_value_create_new_range), (gst_value_subtract_int_range_int), (gst_value_subtract_int_range_int_range), (gst_value_subtract_double_double_range), (gst_value_subtract_double_range_double), (gst_value_subtract_double_range_double_range), (gst_value_subtract_from_list), (gst_value_subtract_list), (gst_value_can_intersect), (gst_value_subtract), (gst_value_can_subtract), (gst_value_register_subtract_func), (_gst_value_initialize): * gst/gstvalue.h: add support for subtracting values from each other. Note that subtracting means subtracting as in set theory. Required for caps stuff above. * testsuite/caps/.cvsignore: * testsuite/caps/Makefile.am: * testsuite/caps/erathostenes.c: (erathostenes), (main): * testsuite/caps/sets.c: (check_caps), (main): * testsuite/caps/subtract.c: (check_caps), (main): add tests for subtraction and equality code.
2004-04-21 03:25:13 +00:00
if (GST_PAD_CAPS (pad) && gst_caps_is_equal (caps, GST_PAD_CAPS (pad))) {
GST_DEBUG ("pad %s:%s already has these caps", GST_DEBUG_PAD_NAME (pad));
return GST_PAD_LINK_OK;
}
link = gst_pad_link_new ();
link->srcpad = GST_PAD_LINK_SRC (pad);
link->sinkpad = GST_PAD_LINK_SINK (pad);
if (!gst_pad_link_ready_for_negotiation (link)) {
GST_DEBUG ("link not ready for negotiating, delaying");
gst_pad_link_free (link);
return GST_PAD_LINK_DELAYED;
}
oldlink = GST_REAL_PAD (pad)->link;
if (oldlink && oldlink->filtercaps) {
link->filtercaps = gst_caps_copy (oldlink->filtercaps);
}
if (link->srcpad == pad) {
link->srccaps = gst_caps_copy (caps);
link->sinkcaps = gst_pad_get_caps (link->sinkpad);
link->srcnotify = FALSE;
} else {
link->srccaps = gst_pad_get_caps (link->srcpad);
link->sinkcaps = gst_caps_copy (caps);
link->sinknotify = FALSE;
}
GST_DEBUG ("trying to link");
ret = gst_pad_link_try (link);
return ret;
}
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
/**
* gst_pad_try_set_caps_nonfixed:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real #GstPad
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* @caps: #GstCaps to set on @pad
*
* Like gst_pad_try_set_caps(), but allows non-fixed caps.
*
* Returns: a #GstPadLinkReturn, like gst_pad_try_set_caps().
*/
GstPadLinkReturn
gst_pad_try_set_caps_nonfixed (GstPad * pad, const GstCaps * caps)
{
GstPadLink *link;
GstPadLink *oldlink;
GstPadLinkReturn ret;
g_return_val_if_fail (GST_IS_REAL_PAD (pad), GST_PAD_LINK_REFUSED);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (!GST_PAD_IS_NEGOTIATING (pad), GST_PAD_LINK_REFUSED);
/* we allow setting caps on non-linked pads. It's ignored */
if (!GST_PAD_PEER (pad)) {
return GST_PAD_LINK_OK;
}
/* we just checked that a peer exists */
g_assert (GST_PAD_LINK_SRC (pad));
g_assert (GST_PAD_LINK_SINK (pad));
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* if the link is already negotiated and the caps are compatible
* with what we're setting, it's trivially OK. */
if (GST_PAD_CAPS (pad)) {
GstCaps *intersection;
intersection = gst_caps_intersect (caps, GST_PAD_CAPS (pad));
if (!gst_caps_is_empty (intersection)) {
gst_caps_free (intersection);
return GST_PAD_LINK_OK;
}
gst_caps_free (intersection);
}
link = gst_pad_link_new ();
link->srcpad = GST_PAD_LINK_SRC (pad);
link->sinkpad = GST_PAD_LINK_SINK (pad);
if (!gst_pad_link_ready_for_negotiation (link)) {
gst_pad_link_free (link);
return GST_PAD_LINK_DELAYED;
}
oldlink = GST_REAL_PAD (pad)->link;
if (oldlink && oldlink->filtercaps) {
link->filtercaps = gst_caps_copy (oldlink->filtercaps);
}
if (link->srcpad == pad) {
link->srccaps = gst_caps_copy (caps);
link->sinkcaps = gst_pad_get_caps (link->sinkpad);
link->srcnotify = FALSE;
} else {
link->srccaps = gst_pad_get_caps (link->srcpad);
link->sinkcaps = gst_caps_copy (caps);
link->sinknotify = FALSE;
}
ret = gst_pad_link_try (link);
return ret;
}
/**
* gst_pad_can_link_filtered:
* @srcpad: the source #GstPad to link.
* @sinkpad: the sink #GstPad to link.
* @filtercaps: the filter #GstCaps.
*
* Checks if the source pad and the sink pad can be linked when constrained
* by the given filter caps. Both @srcpad and @sinkpad must be unlinked.
*
* Returns: TRUE if the pads can be linked, FALSE otherwise.
*/
gboolean
gst_pad_can_link_filtered (GstPad * srcpad, GstPad * sinkpad,
const GstCaps * filtercaps)
{
GstRealPad *realsrc, *realsink;
GstPadLink *link;
/* FIXME This function is gross. It's almost a direct copy of
* gst_pad_link_filtered(). Any decent programmer would attempt
* to merge the two functions, which I will do some day. --ds
*/
/* generic checks */
g_return_val_if_fail (srcpad != NULL, FALSE);
g_return_val_if_fail (GST_IS_PAD (srcpad), FALSE);
g_return_val_if_fail (sinkpad != NULL, FALSE);
g_return_val_if_fail (GST_IS_PAD (sinkpad), FALSE);
GST_CAT_INFO (GST_CAT_PADS, "trying to link %s:%s and %s:%s",
GST_DEBUG_PAD_NAME (srcpad), GST_DEBUG_PAD_NAME (sinkpad));
/* now we need to deal with the real/ghost stuff */
realsrc = GST_PAD_REALIZE (srcpad);
realsink = GST_PAD_REALIZE (sinkpad);
if ((GST_PAD (realsrc) != srcpad) || (GST_PAD (realsink) != sinkpad)) {
GST_CAT_INFO (GST_CAT_PADS, "*actually* linking %s:%s and %s:%s",
GST_DEBUG_PAD_NAME (realsrc), GST_DEBUG_PAD_NAME (realsink));
}
/* FIXME: shouldn't we convert this to g_return_val_if_fail? */
if (GST_RPAD_PEER (realsrc) != NULL) {
GST_CAT_INFO (GST_CAT_PADS, "Real source pad %s:%s has a peer, failed",
GST_DEBUG_PAD_NAME (realsrc));
return FALSE;
}
if (GST_RPAD_PEER (realsink) != NULL) {
GST_CAT_INFO (GST_CAT_PADS, "Real sink pad %s:%s has a peer, failed",
GST_DEBUG_PAD_NAME (realsink));
return FALSE;
}
if (GST_PAD_PARENT (realsrc) == NULL) {
GST_CAT_INFO (GST_CAT_PADS, "Real src pad %s:%s has no parent, failed",
GST_DEBUG_PAD_NAME (realsrc));
return FALSE;
}
if (GST_PAD_PARENT (realsink) == NULL) {
GST_CAT_INFO (GST_CAT_PADS, "Real sink pad %s:%s has no parent, failed",
GST_DEBUG_PAD_NAME (realsrc));
return FALSE;
}
if (!gst_pad_check_schedulers (realsrc, realsink)) {
g_warning ("linking pads with different scheds requires "
"exactly one decoupled element (such as queue)");
return FALSE;
}
g_return_val_if_fail (realsrc != NULL, GST_PAD_LINK_REFUSED);
g_return_val_if_fail (realsink != NULL, GST_PAD_LINK_REFUSED);
link = gst_pad_link_new ();
if (GST_RPAD_DIRECTION (realsrc) == GST_PAD_SRC) {
link->srcpad = GST_PAD (realsrc);
link->sinkpad = GST_PAD (realsink);
} else {
link->srcpad = GST_PAD (realsink);
link->sinkpad = GST_PAD (realsrc);
}
if (GST_RPAD_DIRECTION (link->srcpad) != GST_PAD_SRC) {
GST_CAT_INFO (GST_CAT_PADS,
"Real src pad %s:%s is not a source pad, failed",
GST_DEBUG_PAD_NAME (link->srcpad));
gst_pad_link_free (link);
return FALSE;
}
if (GST_RPAD_DIRECTION (link->sinkpad) != GST_PAD_SINK) {
GST_CAT_INFO (GST_CAT_PADS, "Real sink pad %s:%s is not a sink pad, failed",
GST_DEBUG_PAD_NAME (link->sinkpad));
gst_pad_link_free (link);
return FALSE;
}
link->srccaps = gst_pad_get_caps (link->srcpad);
link->sinkcaps = gst_pad_get_caps (link->sinkpad);
if (filtercaps)
link->filtercaps = gst_caps_copy (filtercaps);
gst_pad_link_intersect (link);
if (gst_caps_is_empty (link->caps)) {
gst_pad_link_free (link);
return FALSE;
}
gst_pad_link_free (link);
return TRUE;
}
/**
* gst_pad_can_link:
* @srcpad: the source #GstPad to link.
* @sinkpad: the sink #GstPad to link.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Checks if the source pad and the sink pad can be linked.
*
* Returns: TRUE if the pads can be linked, FALSE otherwise.
*/
gboolean
gst_pad_can_link (GstPad * srcpad, GstPad * sinkpad)
{
return gst_pad_can_link_filtered (srcpad, sinkpad, NULL);
}
/**
* gst_pad_link_filtered:
* @srcpad: the source #GstPad to link.
* @sinkpad: the sink #GstPad to link.
* @filtercaps: the filter #GstCaps.
*
* Links the source pad and the sink pad, constrained
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* by the given filter caps.
*
* Returns: TRUE if the pads have been linked, FALSE otherwise.
*/
gboolean
gst_pad_link_filtered (GstPad * srcpad, GstPad * sinkpad,
const GstCaps * filtercaps)
{
GstRealPad *realsrc, *realsink;
GstScheduler *src_sched, *sink_sched;
GstPadLink *link;
/* generic checks */
g_return_val_if_fail (srcpad != NULL, FALSE);
g_return_val_if_fail (GST_IS_PAD (srcpad), FALSE);
g_return_val_if_fail (sinkpad != NULL, FALSE);
g_return_val_if_fail (GST_IS_PAD (sinkpad), FALSE);
GST_CAT_INFO (GST_CAT_PADS, "trying to link %s:%s and %s:%s",
GST_DEBUG_PAD_NAME (srcpad), GST_DEBUG_PAD_NAME (sinkpad));
/* now we need to deal with the real/ghost stuff */
realsrc = GST_PAD_REALIZE (srcpad);
realsink = GST_PAD_REALIZE (sinkpad);
if ((GST_PAD (realsrc) != srcpad) || (GST_PAD (realsink) != sinkpad)) {
GST_CAT_INFO (GST_CAT_PADS, "*actually* linking %s:%s and %s:%s",
GST_DEBUG_PAD_NAME (realsrc), GST_DEBUG_PAD_NAME (realsink));
}
/* FIXME: shouldn't we convert this to g_return_val_if_fail? */
if (GST_RPAD_PEER (realsrc) != NULL) {
GST_CAT_INFO (GST_CAT_PADS, "Real source pad %s:%s has a peer, failed",
GST_DEBUG_PAD_NAME (realsrc));
return FALSE;
}
if (GST_RPAD_PEER (realsink) != NULL) {
GST_CAT_INFO (GST_CAT_PADS, "Real sink pad %s:%s has a peer, failed",
GST_DEBUG_PAD_NAME (realsink));
return FALSE;
}
if (GST_PAD_PARENT (realsrc) == NULL) {
GST_CAT_INFO (GST_CAT_PADS, "Real src pad %s:%s has no parent, failed",
GST_DEBUG_PAD_NAME (realsrc));
return FALSE;
}
if (GST_PAD_PARENT (realsink) == NULL) {
GST_CAT_INFO (GST_CAT_PADS, "Real sink pad %s:%s has no parent, failed",
GST_DEBUG_PAD_NAME (realsrc));
return FALSE;
}
if (!gst_pad_check_schedulers (realsrc, realsink)) {
g_warning ("linking pads with different scheds requires "
"exactly one decoupled element (such as queue)");
return FALSE;
}
g_return_val_if_fail (realsrc != NULL, GST_PAD_LINK_REFUSED);
g_return_val_if_fail (realsink != NULL, GST_PAD_LINK_REFUSED);
link = gst_pad_link_new ();
if (GST_RPAD_DIRECTION (realsrc) == GST_PAD_SRC) {
link->srcpad = GST_PAD (realsrc);
link->sinkpad = GST_PAD (realsink);
} else {
link->srcpad = GST_PAD (realsink);
link->sinkpad = GST_PAD (realsrc);
}
if (GST_RPAD_DIRECTION (link->srcpad) != GST_PAD_SRC) {
GST_CAT_INFO (GST_CAT_PADS,
"Real src pad %s:%s is not a source pad, failed",
GST_DEBUG_PAD_NAME (link->srcpad));
gst_pad_link_free (link);
return FALSE;
}
if (GST_RPAD_DIRECTION (link->sinkpad) != GST_PAD_SINK) {
GST_CAT_INFO (GST_CAT_PADS, "Real sink pad %s:%s is not a sink pad, failed",
GST_DEBUG_PAD_NAME (link->sinkpad));
gst_pad_link_free (link);
return FALSE;
}
link->srccaps = gst_pad_get_caps (link->srcpad);
link->sinkcaps = gst_pad_get_caps (link->sinkpad);
if (filtercaps)
link->filtercaps = gst_caps_copy (filtercaps);
if (gst_pad_link_try (link) == GST_PAD_LINK_REFUSED)
return FALSE;
/* fire off a signal to each of the pads telling them
* that they've been linked */
g_signal_emit (G_OBJECT (link->srcpad), gst_real_pad_signals[REAL_LINKED],
0, link->sinkpad);
g_signal_emit (G_OBJECT (link->sinkpad), gst_real_pad_signals[REAL_LINKED],
0, link->srcpad);
src_sched = gst_pad_get_scheduler (GST_PAD (link->srcpad));
sink_sched = gst_pad_get_scheduler (GST_PAD (link->sinkpad));
/* now tell the scheduler */
if (src_sched && src_sched == sink_sched) {
gst_scheduler_pad_link (src_sched,
GST_PAD (link->srcpad), GST_PAD (link->sinkpad));
} else {
GST_CAT_INFO (GST_CAT_PADS,
"not telling link to scheduler %s:%s and %s:%s, %p %p",
GST_DEBUG_PAD_NAME (link->srcpad), GST_DEBUG_PAD_NAME (link->sinkpad),
src_sched, sink_sched);
}
GST_CAT_INFO (GST_CAT_PADS, "linked %s:%s and %s:%s, successful",
GST_DEBUG_PAD_NAME (link->srcpad), GST_DEBUG_PAD_NAME (link->sinkpad));
return TRUE;
}
/**
* gst_pad_link:
* @srcpad: the source #GstPad to link.
* @sinkpad: the sink #GstPad to link.
*
* Links the source pad to the sink pad.
*
* Returns: TRUE if the pad could be linked, FALSE otherwise.
*/
gboolean
gst_pad_link (GstPad * srcpad, GstPad * sinkpad)
{
return gst_pad_link_filtered (srcpad, sinkpad, NULL);
}
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* FIXME 0.9: Remove this */
/**
* gst_pad_set_parent:
* @pad: a #GstPad to set the parent of.
* @parent: the new parent #GstElement.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Sets the parent object of a pad. Deprecated, use gst_object_set_parent()
* instead.
*/
void
gst_pad_set_parent (GstPad * pad, GstElement * parent)
{
g_return_if_fail (GST_IS_PAD (pad));
g_return_if_fail (GST_PAD_PARENT (pad) == NULL);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_if_fail (GST_IS_ELEMENT (parent));
gst_object_set_parent (GST_OBJECT (pad), GST_OBJECT (parent));
}
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* FIXME 0.9: Remove this */
/**
* gst_pad_get_parent:
* @pad: the #GstPad to get the parent of.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Gets the parent object of this pad. Deprecated, use gst_object_get_parent()
* instead.
*
* Returns: the parent #GstElement.
*/
GstElement *
gst_pad_get_parent (GstPad * pad)
{
g_return_val_if_fail (pad != NULL, NULL);
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
return GST_PAD_PARENT (pad);
}
static void
gst_pad_set_pad_template (GstPad * pad, GstPadTemplate * templ)
{
/* this function would need checks if it weren't static */
gst_object_replace ((GstObject **) & pad->padtemplate, (GstObject *) templ);
if (templ) {
gst_object_sink (GST_OBJECT (templ));
g_signal_emit (G_OBJECT (templ),
gst_pad_template_signals[TEMPL_PAD_CREATED], 0, pad);
}
}
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/**
* gst_pad_get_pad_template:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a #GstPad.
*
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* Gets the template for @pad.
*
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* Returns: the #GstPadTemplate from which this pad was instantiated, or %NULL
* if this pad has no template.
*/
GstPadTemplate *
gst_pad_get_pad_template (GstPad * pad)
{
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
return GST_PAD_PAD_TEMPLATE (pad);
}
/**
* gst_pad_get_scheduler:
* @pad: a #GstPad to get the scheduler of.
*
* Gets the scheduler of the pad. Since the pad does not
* have a scheduler of its own, the scheduler of the parent
* is taken. For decoupled pads, the scheduler of the peer
* parent is taken.
*
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* Returns: the #GstScheduler of the pad, or %NULL if there is no parent or the
* parent is not yet in a managing bin.
*/
GstScheduler *
gst_pad_get_scheduler (GstPad * pad)
{
GstScheduler *scheduler = NULL;
GstElement *parent;
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
parent = gst_pad_get_parent (pad);
if (parent) {
if (GST_FLAG_IS_SET (parent, GST_ELEMENT_DECOUPLED)) {
GstRealPad *peer = GST_RPAD_PEER (pad);
if (peer) {
scheduler =
gst_element_get_scheduler (gst_pad_get_parent (GST_PAD (peer)));
}
} else {
scheduler = gst_element_get_scheduler (parent);
}
}
return scheduler;
}
/**
* gst_pad_get_real_parent:
* @pad: a #GstPad to get the real parent of.
*
* Gets the real parent object of this pad. If the pad
* is a ghost pad, the actual owner of the real pad is
* returned, as opposed to #gst_pad_get_parent().
*
* Returns: the parent #GstElement.
*/
GstElement *
gst_pad_get_real_parent (GstPad * pad)
{
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
return GST_PAD_PARENT (GST_PAD (GST_PAD_REALIZE (pad)));
}
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* FIXME 0.9: Make static. */
/**
* gst_pad_add_ghost_pad:
* @pad: a #GstPad to attach the ghost pad to.
* @ghostpad: the ghost #GstPad to to the pad.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Adds a ghost pad to a pad. Private function, will be removed from the API in
* 0.9.
*/
void
gst_pad_add_ghost_pad (GstPad * pad, GstPad * ghostpad)
{
GstRealPad *realpad;
g_return_if_fail (GST_IS_PAD (pad));
g_return_if_fail (GST_IS_GHOST_PAD (ghostpad));
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
/* if we're ghosting a ghost pad, drill down to find the real pad */
realpad = (GstRealPad *) pad;
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
while (GST_IS_GHOST_PAD (realpad))
realpad = GST_GPAD_REALPAD (realpad);
g_return_if_fail (GST_IS_REAL_PAD (realpad));
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
/* will ref the pad template */
GST_GPAD_REALPAD (ghostpad) = realpad;
realpad->ghostpads = g_list_prepend (realpad->ghostpads, ghostpad);
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
gst_pad_set_pad_template (GST_PAD (ghostpad), GST_PAD_PAD_TEMPLATE (pad));
}
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* FIXME 0.9: Make static. */
/**
* gst_pad_remove_ghost_pad:
* @pad: a #GstPad to remove the ghost pad from.
* @ghostpad: the ghost #GstPad to remove from the pad.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Removes a ghost pad from a pad. Private, will be removed from the API in 0.9.
*/
void
gst_pad_remove_ghost_pad (GstPad * pad, GstPad * ghostpad)
{
GstRealPad *realpad;
g_return_if_fail (GST_IS_PAD (pad));
g_return_if_fail (GST_IS_GHOST_PAD (ghostpad));
realpad = GST_PAD_REALIZE (pad);
g_return_if_fail (GST_GPAD_REALPAD (ghostpad) == realpad);
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
gst_pad_set_pad_template (GST_PAD (ghostpad), NULL);
realpad->ghostpads = g_list_remove (realpad->ghostpads, ghostpad);
GST_GPAD_REALPAD (ghostpad) = NULL;
}
/**
* gst_pad_get_ghost_pad_list:
* @pad: a #GstPad to get the ghost pads of.
*
* Gets the ghost pads of this pad.
*
* Returns: a #GList of ghost pads.
*/
GList *
gst_pad_get_ghost_pad_list (GstPad * pad)
{
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
return GST_PAD_REALIZE (pad)->ghostpads;
}
static gboolean
_gst_pad_default_fixate_value (const GValue * value, GValue * dest)
{
GType type = G_VALUE_TYPE (value);
if (gst_value_is_fixed (value))
return TRUE;
if (type == GST_TYPE_INT_RANGE) {
g_value_init (dest, G_TYPE_INT);
g_value_set_int (dest, gst_value_get_int_range_min (value));
} else if (type == GST_TYPE_DOUBLE_RANGE) {
g_value_init (dest, G_TYPE_DOUBLE);
g_value_set_double (dest, gst_value_get_double_range_min (value));
} else if (type == GST_TYPE_LIST) {
gst_value_init_and_copy (dest, gst_value_list_get_value (value, 0));
} else if (type == GST_TYPE_FIXED_LIST) {
gint size, n;
GValue dest_kid = { 0 };
const GValue *kid;
/* check recursively */
g_value_init (dest, GST_TYPE_FIXED_LIST);
size = gst_value_list_get_size (value);
for (n = 0; n < size; n++) {
kid = gst_value_list_get_value (value, n);
if (_gst_pad_default_fixate_value (kid, &dest_kid)) {
gst_value_list_append_value (dest, kid);
} else {
gst_value_list_append_value (dest, &dest_kid);
g_value_unset (&dest_kid);
}
}
} else {
g_critical ("Don't know how to fixate value type %s", g_type_name (type));
}
return FALSE;
}
static gboolean
_gst_pad_default_fixate_foreach (GQuark field_id, GValue * value, gpointer s)
{
GstStructure *structure = (GstStructure *) s;
GValue dest = { 0 };
if (_gst_pad_default_fixate_value (value, &dest))
return TRUE;
gst_structure_id_set_value (structure, field_id, &dest);
g_value_unset (&dest);
return FALSE;
}
static GstCaps *
_gst_pad_default_fixate_func (GstPad * pad, const GstCaps * caps)
{
static GstStaticCaps octetcaps = GST_STATIC_CAPS ("application/octet-stream");
GstStructure *structure;
GstCaps *newcaps;
g_return_val_if_fail (pad != NULL, NULL);
g_return_val_if_fail (caps != NULL, NULL);
g_return_val_if_fail (!gst_caps_is_empty (caps), NULL);
if (gst_caps_is_any (caps)) {
return gst_caps_copy (gst_static_caps_get (&octetcaps));
}
if (caps->structs->len > 1) {
return gst_caps_new_full (gst_structure_copy (gst_caps_get_structure (caps,
0)), NULL);
}
newcaps = gst_caps_copy (caps);
structure = gst_caps_get_structure (newcaps, 0);
gst_structure_foreach (structure, _gst_pad_default_fixate_foreach, structure);
return newcaps;
}
/**
* gst_pad_perform_negotiate:
* @srcpad: the source #GstPad.
* @sinkpad: the sink #GstPad.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Tries to negotiate the pads. See gst_pad_renegotiate() for a brief
* description of caps negotiation.
*
* Returns: TRUE if the pads were succesfully negotiated, FALSE otherwise.
*/
gboolean
gst_pad_perform_negotiate (GstPad * srcpad, GstPad * sinkpad)
{
return GST_PAD_LINK_SUCCESSFUL (gst_pad_renegotiate (srcpad));
}
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
static void
gst_pad_link_unnegotiate (GstPadLink * link)
{
g_return_if_fail (link != NULL);
if (link->caps) {
gst_caps_free (link->caps);
link->caps = NULL;
link->engaged = FALSE;
if (GST_RPAD_LINK (link->srcpad) != link) {
g_warning ("unnegotiating unset link");
} else {
g_object_notify (G_OBJECT (link->srcpad), "caps");
}
if (GST_RPAD_LINK (link->sinkpad) != link) {
g_warning ("unnegotiating unset link");
} else {
g_object_notify (G_OBJECT (link->sinkpad), "caps");
}
}
}
/**
* gst_pad_unnegotiate:
* @pad: pad to unnegotiate
*
* "Unnegotiates" a pad. The currently negotiated caps are cleared and the pad
* needs renegotiation.
*/
void
gst_pad_unnegotiate (GstPad * pad)
{
GstPadLink *link;
g_return_if_fail (GST_IS_PAD (pad));
link = GST_RPAD_LINK (GST_PAD_REALIZE (pad));
if (link)
gst_pad_link_unnegotiate (link);
}
/* returning NULL indicates that the arguments are invalid */
static GstPadLink *
gst_pad_link_prepare (GstPad * srcpad, GstPad * sinkpad,
const GstCaps * filtercaps)
{
GstRealPad *realsrc, *realsink;
GstPadLink *link;
g_return_val_if_fail (GST_IS_PAD (srcpad), NULL);
g_return_val_if_fail (GST_IS_PAD (sinkpad), NULL);
realsrc = GST_PAD_REALIZE (srcpad);
realsink = GST_PAD_REALIZE (sinkpad);
if ((GST_PAD (realsrc) != srcpad) || (GST_PAD (realsink) != sinkpad)) {
GST_CAT_DEBUG (GST_CAT_PADS, "*actually* linking %s:%s and %s:%s",
GST_DEBUG_PAD_NAME (realsrc), GST_DEBUG_PAD_NAME (realsink));
}
g_return_val_if_fail (GST_PAD_PARENT (realsrc) != NULL, NULL);
g_return_val_if_fail (GST_PAD_PARENT (realsink) != NULL, NULL);
if (!gst_pad_check_schedulers (realsrc, realsink)) {
g_warning ("linking pads with different scheds requires "
"exactly one decoupled element (such as queue)");
return NULL;
}
if (GST_RPAD_DIRECTION (realsrc) == GST_RPAD_DIRECTION (realsink)) {
g_warning ("%s:%s and %s:%s are both %s pads, failed",
GST_DEBUG_PAD_NAME (realsrc), GST_DEBUG_PAD_NAME (realsink),
GST_RPAD_DIRECTION (realsrc) == GST_PAD_SRC ? "src" : "sink");
return NULL;
}
link = gst_pad_link_new ();
if (GST_RPAD_DIRECTION (realsrc) == GST_PAD_SRC) {
link->srcpad = GST_PAD (realsrc);
link->sinkpad = GST_PAD (realsink);
} else {
link->srcpad = GST_PAD (realsink);
link->sinkpad = GST_PAD (realsrc);
}
link->srccaps = gst_pad_get_caps (link->srcpad);
link->sinkcaps = gst_pad_get_caps (link->sinkpad);
if (filtercaps)
link->filtercaps = gst_caps_copy (filtercaps);
return link;
}
/**
* gst_pad_try_relink_filtered:
* @srcpad: the source #GstPad to relink.
* @sinkpad: the sink #GstPad to relink.
* @filtercaps: the #GstPad to use as a filter in the relink.
*
* Tries to relink the given source and sink pad, constrained by the given
* capabilities.
*
* Returns: TRUE if the pads were succesfully renegotiated, FALSE otherwise.
*/
gboolean
gst_pad_try_relink_filtered (GstPad * srcpad, GstPad * sinkpad,
const GstCaps * filtercaps)
{
GstPadLink *link;
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
GST_INFO ("trying to relink %" GST_PTR_FORMAT " and %" GST_PTR_FORMAT
" with filtercaps %" GST_PTR_FORMAT, srcpad, sinkpad);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
link = gst_pad_link_prepare (srcpad, sinkpad, filtercaps);
if (!link)
return FALSE;
if (GST_RPAD_PEER (link->srcpad) != (GstRealPad *) link->sinkpad) {
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_warning ("Pads %s:%s and %s:%s were never linked",
GST_DEBUG_PAD_NAME (srcpad), GST_DEBUG_PAD_NAME (sinkpad));
gst_pad_link_free (link);
return FALSE;
}
if (GST_PAD_LINK_FAILED (gst_pad_link_try (link)))
return FALSE;
return TRUE;
}
/**
* gst_pad_relink_filtered:
* @srcpad: the source #GstPad to relink.
* @sinkpad: the sink #GstPad to relink.
* @filtercaps: the #GstPad to use as a filter in the relink.
*
* Relinks the given source and sink pad, constrained by the given
* capabilities. If the relink fails, the pads are unlinked
* and FALSE is returned.
*
* Returns: TRUE if the pads were succesfully relinked, FALSE otherwise.
*/
gboolean
gst_pad_relink_filtered (GstPad * srcpad, GstPad * sinkpad,
const GstCaps * filtercaps)
{
if (gst_pad_try_relink_filtered (srcpad, sinkpad, filtercaps))
return TRUE;
gst_pad_unlink (srcpad, sinkpad);
return FALSE;
}
/**
* gst_pad_proxy_getcaps:
* @pad: a #GstPad to proxy.
*
* Calls gst_pad_get_allowed_caps() for every other pad belonging to the
* same element as @pad, and returns the intersection of the results.
*
* This function is useful as a default getcaps function for an element
* that can handle any stream format, but requires all its pads to have
* the same caps. Two such elements are tee and aggregator.
*
* Returns: the intersection of the other pads' allowed caps.
*/
GstCaps *
gst_pad_proxy_getcaps (GstPad * pad)
{
GstElement *element;
const GList *pads;
GstCaps *caps, *intersected;
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
GST_DEBUG ("proxying getcaps for %s:%s", GST_DEBUG_PAD_NAME (pad));
element = gst_pad_get_parent (pad);
pads = gst_element_get_pad_list (element);
caps = gst_caps_new_any ();
while (pads) {
GstPad *otherpad = GST_PAD (pads->data);
GstCaps *temp;
if (otherpad != pad) {
GstCaps *allowed = gst_pad_get_allowed_caps (otherpad);
temp = gst_caps_intersect (caps, allowed);
gst_caps_free (caps);
gst_caps_free (allowed);
caps = temp;
}
pads = g_list_next (pads);
}
intersected = gst_caps_intersect (caps, gst_pad_get_pad_template_caps (pad));
gst_caps_free (caps);
return intersected;
}
/**
* gst_pad_proxy_pad_link:
* @pad: a #GstPad to proxy from
* @caps: the #GstCaps to link with
*
* Calls gst_pad_try_set_caps() for every other pad belonging to the
* same element as @pad. If gst_pad_try_set_caps() fails on any pad,
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* the proxy link fails. May be used only during negotiation.
*
* Returns: GST_PAD_LINK_OK if sucessful
*/
GstPadLinkReturn
gst_pad_proxy_pad_link (GstPad * pad, const GstCaps * caps)
{
GstElement *element;
const GList *pads;
GstPadLinkReturn ret;
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (GST_IS_PAD (pad), GST_PAD_LINK_REFUSED);
g_return_val_if_fail (caps != NULL, GST_PAD_LINK_REFUSED);
GST_DEBUG ("proxying pad link for %s:%s", GST_DEBUG_PAD_NAME (pad));
element = gst_pad_get_parent (pad);
pads = gst_element_get_pad_list (element);
while (pads) {
GstPad *otherpad = GST_PAD (pads->data);
if (otherpad != pad) {
ret = gst_pad_try_set_caps (otherpad, caps);
if (GST_PAD_LINK_FAILED (ret)) {
return ret;
}
}
pads = g_list_next (pads);
}
return GST_PAD_LINK_OK;
}
/**
* gst_pad_proxy_fixate:
* @pad: a #GstPad to proxy.
* @caps: the #GstCaps to fixate
*
* Implements a default fixate function based on the caps set on the other
* pads in the element. This function should only be used if every pad
* has the same pad template caps.
*
* Returns: a fixated caps, or NULL if caps cannot be fixed
*/
GstCaps *
gst_pad_proxy_fixate (GstPad * pad, const GstCaps * caps)
{
GstElement *element;
const GList *pads;
const GstCaps *othercaps;
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
g_return_val_if_fail (caps != NULL, NULL);
GST_DEBUG ("proxying fixate for %s:%s\n", GST_DEBUG_PAD_NAME (pad));
element = gst_pad_get_parent (pad);
pads = gst_element_get_pad_list (element);
while (pads) {
GstPad *otherpad = GST_PAD (pads->data);
/* FIXME check that each pad has the same pad template caps */
if (otherpad != pad) {
othercaps = gst_pad_get_negotiated_caps (otherpad);
if (othercaps && !gst_caps_is_subset (caps, othercaps)) {
GstCaps *icaps;
icaps = gst_caps_intersect (othercaps, caps);
if (!gst_caps_is_empty (icaps)) {
return icaps;
} else {
gst_caps_free (icaps);
}
}
}
pads = g_list_next (pads);
}
return NULL;
}
/**
* gst_pad_set_explicit_caps:
* @pad: a #GstPad to set the explicit caps of
* @caps: the #GstCaps to set
*
* If a pad has been told to use explicit caps, this function is used
* to set the explicit caps. If @caps is NULL, the explicit caps are
* unset.
*
* This function calls gst_pad_try_set_caps() on the pad. If that
* call fails, GST_ELEMENT_ERROR() is called to indicate a negotiation
* failure.
*
* Returns: TRUE if the caps were set correctly, otherwise FALSE
*/
gboolean
gst_pad_set_explicit_caps (GstPad * pad, const GstCaps * caps)
{
GstPadLinkReturn link_ret;
g_return_val_if_fail (GST_IS_PAD (pad), FALSE);
g_return_val_if_fail (caps == NULL || gst_caps_is_fixed (caps), FALSE);
GST_CAT_DEBUG (GST_CAT_PADS,
"setting explicit caps on %s:%s to %" GST_PTR_FORMAT,
GST_DEBUG_PAD_NAME (pad), caps);
if (caps == NULL) {
GST_CAT_DEBUG (GST_CAT_PADS, "caps is NULL");
gst_caps_replace (&GST_RPAD_EXPLICIT_CAPS (pad), NULL);
return TRUE;
}
gst_caps_replace (&GST_RPAD_EXPLICIT_CAPS (pad), gst_caps_copy (caps));
if (!GST_PAD_IS_LINKED (pad)) {
GST_CAT_DEBUG (GST_CAT_PADS, "pad %s:%s is not linked",
GST_DEBUG_PAD_NAME (pad));
return TRUE;
}
link_ret = gst_pad_try_set_caps (pad, caps);
if (link_ret == GST_PAD_LINK_REFUSED) {
gchar *caps_str = gst_caps_to_string (caps);
GST_ELEMENT_ERROR (gst_pad_get_parent (pad), CORE, PAD, (NULL),
("failed to negotiate (try_set_caps with \"%s\" returned REFUSED)",
caps_str));
g_free (caps_str);
return FALSE;
}
return TRUE;
}
static GstCaps *
gst_pad_explicit_getcaps (GstPad * pad)
{
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
if (GST_RPAD_EXPLICIT_CAPS (pad) == NULL) {
const GstCaps *caps = gst_pad_get_pad_template_caps (pad);
return gst_caps_copy (caps);
}
return gst_caps_copy (GST_RPAD_EXPLICIT_CAPS (pad));
}
static GstPadLinkReturn
gst_pad_explicit_link (GstPad * pad, const GstCaps * caps)
{
g_return_val_if_fail (GST_IS_PAD (pad), GST_PAD_LINK_REFUSED);
g_return_val_if_fail (caps != NULL, GST_PAD_LINK_REFUSED);
if (GST_RPAD_EXPLICIT_CAPS (pad) == NULL) {
return GST_PAD_LINK_DELAYED;
}
return GST_PAD_LINK_OK;
}
/**
* gst_pad_use_explicit_caps:
* @pad: a #GstPad to set to use explicit caps
*
* This function handles negotiation for pads that need to be set
* to particular caps under complete control of the element, based
* on some state in the element. This is often the case with
* decoders and other elements whose caps is determined by the data
* stream.
*
* WARNING: This function is a hack and will be replaced with something
* better in gstreamer-0.9.
*/
void
gst_pad_use_explicit_caps (GstPad * pad)
{
g_return_if_fail (GST_IS_PAD (pad));
gst_pad_set_getcaps_function (pad, gst_pad_explicit_getcaps);
gst_pad_set_link_function (pad, gst_pad_explicit_link);
gst_caps_replace (&GST_RPAD_EXPLICIT_CAPS (pad), NULL);
}
/**
* gst_pad_proxy_link:
* @pad: a #GstPad to proxy to.
* @caps: the #GstCaps to use in proxying.
*
* Proxies the link function to the specified pad.
*
* Returns: TRUE if the peer pad accepted the caps, FALSE otherwise.
*/
GstPadLinkReturn
gst_pad_proxy_link (GstPad * pad, const GstCaps * caps)
{
return gst_pad_try_set_caps (pad, caps);
}
/**
* gst_pad_is_negotiated:
* @pad: a #GstPad to get the negotiation status of
*
* Returns: TRUE if the pad has successfully negotiated caps.
*/
gboolean
gst_pad_is_negotiated (GstPad * pad)
{
g_return_val_if_fail (GST_IS_PAD (pad), FALSE);
if (!(pad = (GstPad *) GST_PAD_REALIZE (pad)))
return FALSE;
if (!GST_RPAD_LINK (pad))
return FALSE;
return (GST_RPAD_LINK (pad)->caps != NULL);
}
/**
* gst_pad_get_negotiated_caps:
* @pad: a #GstPad to get the negotiated capabilites of
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Gets the currently negotiated caps of a pad.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Returns: the currently negotiated caps of a pad, or NULL if the pad isn't
* negotiated.
*/
G_CONST_RETURN GstCaps *
gst_pad_get_negotiated_caps (GstPad * pad)
{
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
if (!(pad = (GstPad *) GST_PAD_REALIZE (pad)))
return NULL;
if (!GST_RPAD_LINK (pad))
return NULL;
return GST_RPAD_LINK (pad)->caps;
}
/**
* gst_pad_get_caps:
* @pad: a #GstPad to get the capabilities of.
*
* Gets the capabilities of this pad.
*
* Returns: the #GstCaps of this pad. This function returns a new caps, so use
* gst_caps_free to get rid of it.
*/
GstCaps *
gst_pad_get_caps (GstPad * pad)
{
GstRealPad *realpad;
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
realpad = GST_PAD_REALIZE (pad);
GST_CAT_DEBUG (GST_CAT_CAPS, "get pad caps of %s:%s (%p)",
GST_DEBUG_PAD_NAME (realpad), realpad);
if (GST_PAD_IS_DISPATCHING (realpad))
GST_CAT_DEBUG (GST_CAT_CAPS,
"pad %s:%s is already dispatching -- looking for a template",
GST_DEBUG_PAD_NAME (realpad));
if (GST_RPAD_GETCAPSFUNC (realpad) && !GST_PAD_IS_DISPATCHING (realpad)) {
GstCaps *caps;
GST_CAT_DEBUG (GST_CAT_CAPS, "dispatching to pad getcaps function");
GST_FLAG_SET (realpad, GST_PAD_DISPATCHING);
caps = GST_RPAD_GETCAPSFUNC (realpad) (GST_PAD (realpad));
GST_FLAG_UNSET (realpad, GST_PAD_DISPATCHING);
if (caps == NULL) {
g_critical ("pad %s:%s returned NULL caps from getcaps function\n",
GST_DEBUG_PAD_NAME (realpad));
} else {
#ifndef G_DISABLE_ASSERT
/* check that the returned caps are a real subset of the template caps */
if (GST_PAD_PAD_TEMPLATE (realpad)) {
const GstCaps *templ_caps =
GST_PAD_TEMPLATE_CAPS (GST_PAD_PAD_TEMPLATE (realpad));
if (!gst_caps_is_subset (caps, templ_caps)) {
GstCaps *temp;
GST_CAT_ERROR_OBJECT (GST_CAT_CAPS, pad,
"pad returned caps %" GST_PTR_FORMAT
" which are not a real subset of its template caps %"
GST_PTR_FORMAT, caps, templ_caps);
g_warning
("pad %s:%s returned caps that are not a real subset of its template caps",
GST_DEBUG_PAD_NAME (realpad));
temp = gst_caps_intersect (templ_caps, caps);
gst_caps_free (caps);
caps = temp;
}
}
#endif
return caps;
}
}
if (GST_PAD_PAD_TEMPLATE (realpad)) {
GstPadTemplate *templ = GST_PAD_PAD_TEMPLATE (realpad);
const GstCaps *caps;
caps = GST_PAD_TEMPLATE_CAPS (templ);
GST_CAT_DEBUG (GST_CAT_CAPS,
"using pad template %p with caps %" GST_PTR_FORMAT, templ, caps);
#if 0
/* FIXME we should enable something like this someday, but this is
* a bit buggy */
if (!gst_caps_is_fixed (caps)) {
g_warning
("pad %s:%s (%p) has no getcaps function and the pad template returns non-fixed caps. Element is probably broken.\n",
GST_DEBUG_PAD_NAME (realpad), realpad);
}
#endif
return gst_caps_copy (GST_PAD_TEMPLATE_CAPS (templ));
}
GST_CAT_DEBUG (GST_CAT_CAPS, "pad has no caps");
#if 0
/* FIXME enable */
g_warning ("pad %s:%s (%p) has no pad template\n",
GST_DEBUG_PAD_NAME (realpad), realpad);
#endif
return gst_caps_new_any ();
}
/**
* gst_pad_get_pad_template_caps:
* @pad: a #GstPad to get the template capabilities from.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Gets the capabilities for @pad's template.
*
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
* Returns: the #GstCaps of this pad template. If you intend to keep a reference
* on the caps, make a copy (see gst_caps_copy ()).
*/
const GstCaps *
gst_pad_get_pad_template_caps (GstPad * pad)
{
static GstStaticCaps anycaps = GST_STATIC_CAPS ("ANY");
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
if (GST_PAD_PAD_TEMPLATE (pad))
return GST_PAD_TEMPLATE_CAPS (GST_PAD_PAD_TEMPLATE (pad));
#if 0
/* FIXME this should be enabled some day */
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* wingo: why? mail the list during 0.9 when you find this :) */
g_warning ("pad %s:%s (%p) has no pad template\n",
GST_DEBUG_PAD_NAME (realpad), realpad);
#endif
return gst_static_caps_get (&anycaps);
}
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* FIXME 0.9: This function should probably die, or at least be renamed to
* get_caps_by_format. */
/**
* gst_pad_template_get_caps_by_name:
* @templ: a #GstPadTemplate to get the capabilities of.
* @name: the name of the capability to get.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Gets the capability with the given name from @templ.
*
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
* Returns: the #GstCaps of this pad template, or NULL if not found. If you
* intend to keep a reference on the caps, make a copy (see gst_caps_copy ()).
*/
const GstCaps *
gst_pad_template_get_caps_by_name (GstPadTemplate * templ, const gchar * name)
{
GstCaps *caps;
g_return_val_if_fail (templ != NULL, NULL);
caps = GST_PAD_TEMPLATE_CAPS (templ);
if (!caps)
return NULL;
/* FIXME */
return NULL;
}
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* FIXME 0.9: What good is this if it only works for already-negotiated pads? */
/**
* gst_pad_check_compatibility:
* @srcpad: the source #GstPad to check.
* @sinkpad: the sink #GstPad to check against.
*
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* Checks if two pads have compatible capabilities. If neither one has yet been
* negotiated, returns TRUE for no good reason.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Returns: TRUE if they are compatible or if the capabilities could not be
* checked, FALSE if the capabilities are not compatible.
*/
gboolean
gst_pad_check_compatibility (GstPad * srcpad, GstPad * sinkpad)
{
g_return_val_if_fail (GST_IS_PAD (srcpad), FALSE);
g_return_val_if_fail (GST_IS_PAD (sinkpad), FALSE);
if (GST_PAD_CAPS (srcpad) && GST_PAD_CAPS (sinkpad)) {
if (!gst_caps_is_always_compatible (GST_PAD_CAPS (srcpad),
GST_PAD_CAPS (sinkpad))) {
return FALSE;
} else {
return TRUE;
}
} else {
GST_CAT_DEBUG (GST_CAT_PADS,
"could not check capabilities of pads (%s:%s) and (%s:%s) %p %p",
GST_DEBUG_PAD_NAME (srcpad), GST_DEBUG_PAD_NAME (sinkpad),
GST_PAD_CAPS (srcpad), GST_PAD_CAPS (sinkpad));
return TRUE;
}
}
/**
* gst_pad_get_peer:
* @pad: a #GstPad to get the peer of.
*
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* Gets the peer of @pad.
*
* Returns: the peer #GstPad.
*/
GstPad *
gst_pad_get_peer (GstPad * pad)
{
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
return GST_PAD (GST_PAD_PEER (pad));
}
/**
* gst_pad_get_allowed_caps:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a real #GstPad.
*
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* Gets the capabilities of the allowed media types that can flow through @pad.
* The caller must free the resulting caps.
*
* Returns: the allowed #GstCaps of the pad link. Free the caps when
* you no longer need it.
*/
GstCaps *
gst_pad_get_allowed_caps (GstPad * pad)
{
const GstCaps *mycaps;
GstCaps *caps;
GstCaps *peercaps;
GstCaps *icaps;
GstPadLink *link;
g_return_val_if_fail (GST_IS_REAL_PAD (pad), NULL);
GST_CAT_DEBUG (GST_CAT_PROPERTIES, "%s:%s: getting allowed caps",
GST_DEBUG_PAD_NAME (pad));
mycaps = gst_pad_get_pad_template_caps (pad);
if (GST_RPAD_PEER (pad) == NULL) {
GST_CAT_DEBUG (GST_CAT_PROPERTIES, "%s:%s: no peer, returning template",
GST_DEBUG_PAD_NAME (pad));
return gst_caps_copy (mycaps);
}
peercaps = gst_pad_get_caps (GST_PAD_PEER (pad));
caps = gst_caps_intersect (mycaps, peercaps);
gst_caps_free (peercaps);
link = GST_RPAD_LINK (pad);
if (link->filtercaps) {
icaps = gst_caps_intersect (caps, link->filtercaps);
gst_caps_free (caps);
GST_CAT_DEBUG (GST_CAT_PROPERTIES,
"%s:%s: returning filtered intersection with peer",
GST_DEBUG_PAD_NAME (pad));
return icaps;
} else {
GST_CAT_DEBUG (GST_CAT_PROPERTIES,
"%s:%s: returning unfiltered intersection with peer",
GST_DEBUG_PAD_NAME (pad));
return caps;
}
}
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
/**
* gst_pad_caps_change_notify:
* @pad: a #GstPad
*
* Called to indicate that the return value of @pad's getcaps function may have
* changed, and that a renegotiation is suggested.
*/
void
gst_pad_caps_change_notify (GstPad * pad)
{
}
/**
* gst_pad_recover_caps_error:
* @pad: a #GstPad that had a failed capsnego
* @allowed: possible caps for the link
*
* Attempt to recover from a failed caps negotiation. This function
* is typically called by a plugin that exhausted its list of caps
* and wants the application to resolve the issue. The application
* should connect to the pad's caps_nego_failed signal and should
* resolve the issue by connecting another element for example.
*
* Returns: TRUE when the issue was resolved, dumps detailed information
* on the console and returns FALSE otherwise.
*/
gboolean
gst_pad_recover_caps_error (GstPad * pad, const GstCaps * allowed)
{
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* FIXME */
return FALSE;
}
/**
* gst_pad_alloc_buffer:
* @pad: a source #GstPad
* @offset: the offset of the new buffer in the stream
* @size: the size of the new buffer
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Allocates a new, empty buffer optimized to push to pad @pad. This
* function only works if @pad is a source pad.
*
* Returns: a new, empty #GstBuffer, or NULL if there is an error
*/
GstBuffer *
gst_pad_alloc_buffer (GstPad * pad, guint64 offset, gint size)
{
GstRealPad *peer;
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
g_return_val_if_fail (GST_PAD_IS_SRC (pad), NULL);
peer = GST_RPAD_PEER (pad);
while (peer) {
if (peer->bufferallocfunc) {
GstBuffer *ret;
GST_CAT_DEBUG (GST_CAT_BUFFER, "(%s:%s): getting buffer",
GST_DEBUG_PAD_NAME (pad));
GST_CAT_DEBUG (GST_CAT_PADS,
"calling bufferallocfunc &%s (@%p) of peer pad %s:%s",
GST_DEBUG_FUNCPTR_NAME (peer->bufferallocfunc),
&peer->bufferallocfunc, GST_DEBUG_PAD_NAME (((GstPad *) peer)));
ret = (peer->bufferallocfunc) (GST_PAD (peer), offset, size);
if (ret)
return ret;
else
break;
} else {
/*
there is a peer but it doesn't have a bufferallocfunc.
Try to proxy the request if GST_ELEMENT_WORK_IN_PLACE
is setted (try only when there is a single src pad)
*/
GstElement *parent = GST_PAD_PARENT (peer);
if (parent == NULL)
break;
if (GST_FLAG_IS_SET (parent, GST_ELEMENT_WORK_IN_PLACE)) {
/* get the src pad. don't proxy if src pad numbers != 1 */
const GList *pads;
GstPad *src = NULL;
gint src_pad_count = 0;
for (pads = gst_element_get_pad_list (parent);
pads != NULL; pads = pads->next) {
GstPad *t = GST_PAD (pads->data);
if (GST_PAD_DIRECTION (t) == GST_PAD_SRC) {
src = t;
src_pad_count++;
}
}
if ((src == NULL) || (src_pad_count != 1))
break;
peer = GST_RPAD_PEER (src);
} else
break;
}
}
return gst_buffer_new_and_alloc (size);
}
static void
gst_real_pad_dispose (GObject * object)
{
GstPad *pad = GST_PAD (object);
/* No linked pad can ever be disposed.
* It has to have a parent to be linked
* and a parent would hold a reference */
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
/* FIXME: what about if g_object_dispose is explicitly called on the pad? Is
that legal? otherwise we could assert GST_OBJECT_PARENT (pad) == NULL as
well... */
g_assert (GST_PAD_PEER (pad) == NULL);
GST_CAT_DEBUG (GST_CAT_REFCOUNTING, "dispose %s:%s",
GST_DEBUG_PAD_NAME (pad));
/* we destroy the ghostpads, because they are nothing without the real pad */
if (GST_REAL_PAD (pad)->ghostpads) {
GList *orig, *ghostpads;
orig = ghostpads = g_list_copy (GST_REAL_PAD (pad)->ghostpads);
while (ghostpads) {
GstPad *ghostpad = GST_PAD (ghostpads->data);
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
if (GST_IS_ELEMENT (GST_OBJECT_PARENT (ghostpad))) {
GstElement *parent = GST_ELEMENT (GST_OBJECT_PARENT (ghostpad));
GST_CAT_DEBUG (GST_CAT_REFCOUNTING,
"removing ghost pad from element '%s'", GST_OBJECT_NAME (parent));
gst_element_remove_pad (parent, ghostpad);
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
} else {
/* handle the case where we have some floating ghost pad that was never
added to an element */
g_object_set (ghostpad, "real-pad", NULL, NULL);
}
ghostpads = g_list_next (ghostpads);
}
g_list_free (orig);
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
/* as the ghost pads are removed, they remove themselves from ->ghostpads.
So it should be empty now. Let's assert that. */
g_assert (GST_REAL_PAD (pad)->ghostpads == NULL);
}
if (GST_IS_ELEMENT (GST_OBJECT_PARENT (pad))) {
GST_CAT_DEBUG (GST_CAT_REFCOUNTING, "removing pad from element '%s'",
GST_OBJECT_NAME (GST_OBJECT (GST_ELEMENT (GST_OBJECT_PARENT (pad)))));
gst_element_remove_pad (GST_ELEMENT (GST_OBJECT_PARENT (pad)), pad);
}
if (GST_RPAD_EXPLICIT_CAPS (pad)) {
GST_ERROR_OBJECT (pad, "still explicit caps %" GST_PTR_FORMAT " set",
GST_RPAD_EXPLICIT_CAPS (pad));
g_warning ("pad %p has still explicit caps set", pad);
gst_caps_replace (&GST_RPAD_EXPLICIT_CAPS (pad), NULL);
}
G_OBJECT_CLASS (real_pad_parent_class)->dispose (object);
}
#ifndef GST_DISABLE_LOADSAVE
/* FIXME: why isn't this on a GstElement ? */
/**
* gst_pad_load_and_link:
* @self: an #xmlNodePtr to read the description from.
* @parent: the #GstObject element that owns the pad.
*
* Reads the pad definition from the XML node and links the given pad
* in the element to a pad of an element up in the hierarchy.
*/
void
gst_pad_load_and_link (xmlNodePtr self, GstObject * parent)
{
xmlNodePtr field = self->xmlChildrenNode;
GstPad *pad = NULL, *targetpad;
gchar *peer = NULL;
gchar **split;
GstElement *target;
GstObject *grandparent;
gchar *name = NULL;
while (field) {
if (!strcmp ((char *) field->name, "name")) {
name = (char *) xmlNodeGetContent (field);
pad = gst_element_get_pad (GST_ELEMENT (parent), name);
g_free (name);
} else if (!strcmp ((char *) field->name, "peer")) {
peer = (char *) xmlNodeGetContent (field);
}
field = field->next;
}
g_return_if_fail (pad != NULL);
if (peer == NULL)
return;
split = g_strsplit (peer, ".", 2);
if (split[0] == NULL || split[1] == NULL) {
GST_CAT_DEBUG (GST_CAT_XML,
"Could not parse peer '%s' for pad %s:%s, leaving unlinked",
peer, GST_DEBUG_PAD_NAME (pad));
g_free (peer);
return;
}
g_free (peer);
g_return_if_fail (split[0] != NULL);
g_return_if_fail (split[1] != NULL);
grandparent = gst_object_get_parent (parent);
if (grandparent && GST_IS_BIN (grandparent)) {
target = gst_bin_get_by_name_recurse_up (GST_BIN (grandparent), split[0]);
} else
goto cleanup;
if (target == NULL)
goto cleanup;
targetpad = gst_element_get_pad (target, split[1]);
if (targetpad == NULL)
goto cleanup;
gst_pad_link (pad, targetpad);
cleanup:
g_strfreev (split);
}
/**
* gst_pad_save_thyself:
* @pad: a #GstPad to save.
* @parent: the parent #xmlNodePtr to save the description in.
*
* Saves the pad into an xml representation.
*
* Returns: the #xmlNodePtr representation of the pad.
*/
static xmlNodePtr
gst_pad_save_thyself (GstObject * object, xmlNodePtr parent)
{
GstRealPad *realpad;
GstPad *peer;
g_return_val_if_fail (GST_IS_REAL_PAD (object), NULL);
realpad = GST_REAL_PAD (object);
xmlNewChild (parent, NULL, (guchar *) "name",
(guchar *) GST_PAD_NAME (realpad));
if (GST_RPAD_PEER (realpad) != NULL) {
gchar *content;
peer = GST_PAD (GST_RPAD_PEER (realpad));
/* first check to see if the peer's parent's parent is the same */
/* we just save it off */
content = g_strdup_printf ("%s.%s",
GST_OBJECT_NAME (GST_PAD_PARENT (peer)), GST_PAD_NAME (peer));
xmlNewChild (parent, NULL, (guchar *) "peer", (guchar *) content);
g_free (content);
} else
xmlNewChild (parent, NULL, (guchar *) "peer", (guchar *) "");
return parent;
}
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
/* FIXME: shouldn't it be gst_pad_ghost_* ?
* dunno -- wingo 7 feb 2004
*/
/**
* gst_ghost_pad_save_thyself:
* @pad: a ghost #GstPad to save.
* @parent: the parent #xmlNodePtr to save the description in.
*
* Saves the ghost pad into an xml representation.
*
* Returns: the #xmlNodePtr representation of the pad.
*/
xmlNodePtr
gst_ghost_pad_save_thyself (GstPad * pad, xmlNodePtr parent)
{
xmlNodePtr self;
g_return_val_if_fail (GST_IS_GHOST_PAD (pad), NULL);
self = xmlNewChild (parent, NULL, (guchar *) "ghostpad", NULL);
xmlNewChild (self, NULL, (guchar *) "name", (guchar *) GST_PAD_NAME (pad));
xmlNewChild (self, NULL, (guchar *) "parent",
(guchar *) GST_OBJECT_NAME (GST_PAD_PARENT (pad)));
/* FIXME FIXME FIXME! */
return self;
}
#endif /* GST_DISABLE_LOADSAVE */
static GstData *
_invent_event (GstPad * pad, GstBuffer * buffer)
{
GstEvent *event;
GstEventType event_type;
guint64 offset;
if (GST_BUFFER_OFFSET_IS_VALID (buffer))
event_type = GST_FORMAT_DEFAULT;
else
event_type = GST_FORMAT_UNDEFINED;
offset = GST_BUFFER_OFFSET (buffer);
if (GST_BUFFER_TIMESTAMP_IS_VALID (buffer)) {
GstClockTime timestamp = GST_BUFFER_TIMESTAMP (buffer);
event = gst_event_new_discontinuous (TRUE,
GST_FORMAT_TIME, timestamp, event_type, offset, GST_FORMAT_UNDEFINED);
GST_CAT_WARNING (GST_CAT_SCHEDULING,
"needed to invent a DISCONT %p (time %" G_GUINT64_FORMAT
") for %s:%s => %s:%s", event, timestamp,
GST_DEBUG_PAD_NAME (GST_PAD_PEER (pad)), GST_DEBUG_PAD_NAME (pad));
} else {
event = gst_event_new_discontinuous (TRUE,
event_type, offset, GST_FORMAT_UNDEFINED);
GST_CAT_WARNING (GST_CAT_SCHEDULING,
"needed to invent a DISCONT %p (no time) for %s:%s => %s:%s", event,
GST_DEBUG_PAD_NAME (GST_PAD_PEER (pad)), GST_DEBUG_PAD_NAME (pad));
}
return GST_DATA (event);
}
/**
* gst_pad_push:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a source #GstPad.
* @data: the #GstData to push.
*
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* Pushes a buffer or an event to the peer of @pad. @pad must be linked. May
* only be called by @pad's parent.
*/
void
gst_pad_push (GstPad * pad, GstData * data)
{
GstRealPad *peer;
g_return_if_fail (GST_IS_PAD (pad));
g_return_if_fail (GST_PAD_DIRECTION (pad) == GST_PAD_SRC);
g_return_if_fail (!GST_FLAG_IS_SET (GST_PAD_REALIZE (pad),
GST_RPAD_IN_GETFUNC));
g_return_if_fail (data != NULL);
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
DEBUG_DATA (pad, data, "gst_pad_push");
if (!gst_probe_dispatcher_dispatch (&(GST_REAL_PAD (pad)->probedisp), &data)) {
GST_CAT_LOG_OBJECT (GST_CAT_SCHEDULING, pad,
"not pushing data %p, blocked by probe", data);
gst_data_unref (data);
return;
}
if (!GST_PAD_IS_LINKED (pad)) {
GST_CAT_LOG_OBJECT (GST_CAT_SCHEDULING, pad,
"not pushing data %p as pad is unconnected", data);
gst_data_unref (data);
return;
}
if (GST_IS_BUFFER (data) && !gst_pad_is_negotiated (pad)) {
g_warning ("pushing data on non-negotiated pad %s:%s, not allowed.",
GST_DEBUG_PAD_NAME (pad));
gst_data_unref (data);
return;
}
GST_CAT_LOG_OBJECT (GST_CAT_SCHEDULING, pad, "pushing");
peer = GST_RPAD_PEER (pad);
if (!peer) {
g_warning ("push on pad %s:%s but it is unlinked",
GST_DEBUG_PAD_NAME (pad));
} else {
if (!GST_IS_EVENT (data) && !GST_PAD_IS_ACTIVE (peer)) {
g_warning ("push on peer of pad %s:%s but peer is not active",
GST_DEBUG_PAD_NAME (pad));
return;
}
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
if (peer->chainhandler) {
if (data) {
if (!gst_probe_dispatcher_dispatch (&peer->probedisp, &data)) {
GST_CAT_LOG_OBJECT (GST_CAT_SCHEDULING, pad,
"not pushing data %p, blocked by probe", data);
gst_data_unref (data);
return;
}
GST_CAT_LOG_OBJECT (GST_CAT_SCHEDULING, pad,
"calling chainhandler &%s of peer pad %s:%s",
GST_DEBUG_FUNCPTR_NAME (peer->chainhandler),
GST_DEBUG_PAD_NAME (GST_PAD (peer)));
(peer->chainhandler) (GST_PAD (peer), data);
return;
} else {
g_warning ("trying to push a NULL buffer on pad %s:%s",
GST_DEBUG_PAD_NAME (peer));
return;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
}
} else {
g_warning ("internal error: push on pad %s:%s but it has no chainhandler",
GST_DEBUG_PAD_NAME (peer));
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
}
}
/* clean up the mess here */
if (data != NULL)
gst_data_unref (data);
}
/**
* gst_pad_pull:
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @pad: a sink #GstPad.
*
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* Pulls an event or a buffer from the peer pad. May only be called by @pad's
* parent.
*
* Returns: a new #GstData from the peer pad.
*/
GstData *
gst_pad_pull (GstPad * pad)
{
GstRealPad *peer;
GstData *data;
g_return_val_if_fail (GST_PAD_DIRECTION (pad) == GST_PAD_SINK,
GST_DATA (gst_event_new (GST_EVENT_INTERRUPT)));
g_return_val_if_fail (!GST_FLAG_IS_SET (GST_PAD_REALIZE (pad),
GST_RPAD_IN_CHAINFUNC),
GST_DATA (gst_event_new (GST_EVENT_INTERRUPT)));
peer = GST_RPAD_PEER (pad);
if (!peer) {
GST_ELEMENT_ERROR (GST_PAD_PARENT (pad), CORE, PAD, (NULL),
("pull on pad %s:%s but it was unlinked", GST_DEBUG_PAD_NAME (pad)));
} else {
restart:
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
if (peer->gethandler) {
GstPadLink *link = GST_RPAD_LINK (pad);
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GST_CAT_LOG_OBJECT (GST_CAT_SCHEDULING, pad,
"calling gethandler %s of peer pad %s:%s",
GST_DEBUG_FUNCPTR_NAME (peer->gethandler), GST_DEBUG_PAD_NAME (peer));
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
if (link->temp_store) {
g_assert (link->engaged);
GST_DEBUG ("moving temp_store %p to data", link->temp_store);
data = link->temp_store;
link->temp_store = NULL;
} else {
data = (peer->gethandler) (GST_PAD (peer));
/* refetch - we might have been relinked */
link = GST_RPAD_LINK (pad);
peer = GST_RPAD_PEER (pad);
if (!peer) {
/* Our peer disa-peer-ed ;) */
goto int_out;
}
}
if (data) {
if (!link->engaged) {
g_assert (link->temp_store == NULL);
if (GST_IS_BUFFER (data)) {
GST_DEBUG ("moving data buffer %p back to temp_store", data);
link->temp_store = data;
link->engaged = TRUE;
data = _invent_event (pad, GST_BUFFER (data));
} else if (GST_IS_EVENT (data) &&
GST_EVENT_TYPE (data) == GST_EVENT_DISCONTINUOUS &&
GST_EVENT_DISCONT_NEW_MEDIA (data)) {
link->engaged = TRUE;
GST_CAT_LOG (GST_CAT_SCHEDULING,
"link engaged by discont event %p for pad %s:%s", data,
GST_DEBUG_PAD_NAME (pad));
}
}
GST_DEBUG ("calling gst_probe_dispatcher_dispatch on data %p", data);
if (!gst_probe_dispatcher_dispatch (&peer->probedisp, &data)) {
GST_CAT_LOG_OBJECT (GST_CAT_SCHEDULING, pad,
"not returning pulled data %p, blocked by probe", data);
gst_data_unref (data);
goto restart;
}
DEBUG_DATA (pad, data, "gst_pad_pull returned");
return data;
}
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
/* no null buffers allowed */
GST_ELEMENT_ERROR (GST_PAD_PARENT (pad), CORE, PAD, (NULL),
("NULL buffer during pull on %s:%s", GST_DEBUG_PAD_NAME (pad)));
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
} else {
GST_ELEMENT_ERROR (GST_PAD_PARENT (pad), CORE, PAD, (NULL),
("pull on pad %s:%s but the peer pad %s:%s has no gethandler",
GST_DEBUG_PAD_NAME (pad), GST_DEBUG_PAD_NAME (peer)));
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
}
}
int_out:
data = GST_DATA (gst_event_new (GST_EVENT_INTERRUPT));
DEBUG_DATA (pad, data, "gst_pad_pull returning INTERRUPT");
return data;
}
2004-04-04 02:36:12 +00:00
GstData *
gst_pad_collect_array (GstScheduler * scheduler, GstPad ** selected,
GstPad ** padlist)
{
GstSchedulerClass *klass = GST_SCHEDULER_GET_CLASS (scheduler);
if (!GST_FLAG_IS_SET (scheduler, GST_SCHEDULER_FLAG_NEW_API) ||
!klass->pad_select) {
/* better randomness? */
if (selected)
*selected = padlist[0];
return gst_pad_pull (padlist[0]);
} else {
GstPad *select;
return klass->pad_select (scheduler, selected ? selected : &select,
padlist);
}
}
/**
2004-04-04 02:36:12 +00:00
* gst_pad_collectv:
* @selected: set to the pad the buffer comes from if not NULL
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* @padlist: a #GList of sink pads.
*
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* Waits for a buffer on any of the list of pads. Each #GstPad in @padlist must
2004-04-04 02:36:12 +00:00
* belong to the same element and be owned by the caller.
*
2004-04-04 02:36:12 +00:00
* Returns: the #GstData that was available
*/
2004-04-04 02:36:12 +00:00
GstData *
gst_pad_collectv (GstPad ** selected, const GList * padlist)
{
/* need to use alloca here because we must not leak data */
GstPad **pads;
GstPad *test, *_tmp;
2004-04-04 02:36:12 +00:00
GstElement *element = NULL;
int i = 0;
g_return_val_if_fail (padlist != NULL, NULL);
pads = g_malloc (sizeof (gpointer) * (g_list_length ((GList *) padlist) + 1));
2004-04-04 02:36:12 +00:00
for (; padlist; padlist = g_list_next (padlist)) {
test = GST_PAD (padlist->data);
g_return_val_if_fail (GST_IS_PAD (test), NULL);
g_return_val_if_fail (GST_PAD_IS_SINK (test), NULL);
if (element) {
g_return_val_if_fail (element == gst_pad_get_parent (test), NULL);
} else {
element = gst_pad_get_parent (test);
}
pads[i++] = test;
}
pads[i] = NULL;
_tmp = gst_pad_collect_array (GST_ELEMENT_SCHED (element), selected, pads);
g_free (pads);
return _tmp;
}
/**
2004-04-04 02:36:12 +00:00
* gst_pad_collect:
* @selected: set to the pad the buffer comes from if not NULL
* @pad: first pad
* @...: more sink pads.
*
* Waits for a buffer on the given set of pads.
*
2004-04-04 02:36:12 +00:00
* Returns: the #GstData that was available.
*/
2004-04-04 02:36:12 +00:00
GstData *
gst_pad_collect (GstPad ** selected, GstPad * pad, ...)
{
2004-04-04 02:36:12 +00:00
GstData *result;
va_list var_args;
2004-04-04 02:36:12 +00:00
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
va_start (var_args, pad);
2004-04-04 02:36:12 +00:00
result = gst_pad_collect_valist (selected, pad, var_args);
va_end (var_args);
return result;
}
2004-04-04 02:36:12 +00:00
/**
* gst_pad_collect_valist:
* @selected: set to the pad the buffer comes from if not NULL
* @pad: first pad
* @...: more sink pads.
*
* Waits for a buffer on the given set of pads.
*
* Returns: the #GstData that was available.
*/
GstData *
gst_pad_collect_valist (GstPad ** selected, GstPad * pad, va_list var_args)
{
GstPad **padlist;
GstPad *_tmp;
2004-04-04 02:36:12 +00:00
GstElement *element = NULL;
gint i = 0, maxlength;
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
element = gst_pad_get_parent (pad);
maxlength = element->numsinkpads;
/* can we make this list a bit smaller than this upper limit? */
padlist = g_malloc (sizeof (gpointer) * (maxlength + 1));
2004-04-04 02:36:12 +00:00
while (pad) {
g_return_val_if_fail (i < maxlength, NULL);
g_return_val_if_fail (element == gst_pad_get_parent (pad), NULL);
2004-04-04 02:36:12 +00:00
padlist[i++] = pad;
pad = va_arg (var_args, GstPad *);
}
padlist[i] = NULL;
_tmp = gst_pad_collect_array (GST_ELEMENT_SCHED (element), selected, padlist);
g_free (padlist);
return _tmp;
2004-04-04 02:36:12 +00:00
}
/**
* gst_pad_selectv:
* @padlist: a #GList of sink pads.
*
* Waits for a buffer on any of the list of pads. Each #GstPad in @padlist must
* be owned by the calling code.
*
* Returns: the #GstPad that has a buffer available.
* Use #gst_pad_pull() to get the buffer.
*/
GstPad *
gst_pad_selectv (GList * padlist)
{
return NULL;
}
/**
* gst_pad_select_valist:
* @pad: a first #GstPad to perform the select on.
* @varargs: A va_list of more pads to select on.
*
* Waits for a buffer on the given set of pads.
*
* Returns: the #GstPad that has a buffer available.
* Use #gst_pad_pull() to get the buffer.
*/
GstPad *
gst_pad_select_valist (GstPad * pad, va_list var_args)
{
GstPad *result;
GList *padlist = NULL;
if (pad == NULL)
return NULL;
while (pad) {
padlist = g_list_prepend (padlist, pad);
pad = va_arg (var_args, GstPad *);
}
result = gst_pad_selectv (padlist);
g_list_free (padlist);
return result;
}
2004-04-04 02:36:12 +00:00
/**
* gst_pad_select:
* @pad: a first sink #GstPad to perform the select on.
* @...: A NULL-terminated list of more pads to select on.
*
* Waits for a buffer on the given set of pads.
*
* Returns: the #GstPad that has a buffer available.
* Use #gst_pad_pull() to get the buffer.
*/
GstPad *
gst_pad_select (GstPad * pad, ...)
{
GstPad *result;
va_list var_args;
if (pad == NULL)
return NULL;
va_start (var_args, pad);
result = gst_pad_select_valist (pad, var_args);
va_end (var_args);
return result;
}
/************************************************************************
*
* templates
*
*/
static void gst_pad_template_class_init (GstPadTemplateClass * klass);
static void gst_pad_template_init (GstPadTemplate * templ);
static void gst_pad_template_dispose (GObject * object);
GType
gst_pad_template_get_type (void)
{
static GType padtemplate_type = 0;
if (!padtemplate_type) {
static const GTypeInfo padtemplate_info = {
sizeof (GstPadTemplateClass), NULL, NULL,
(GClassInitFunc) gst_pad_template_class_init, NULL, NULL,
sizeof (GstPadTemplate),
0,
(GInstanceInitFunc) gst_pad_template_init, NULL
};
padtemplate_type =
g_type_register_static (GST_TYPE_OBJECT, "GstPadTemplate",
&padtemplate_info, 0);
}
return padtemplate_type;
}
static void
gst_pad_template_class_init (GstPadTemplateClass * klass)
{
GObjectClass *gobject_class;
GstObjectClass *gstobject_class;
gobject_class = (GObjectClass *) klass;
gstobject_class = (GstObjectClass *) klass;
padtemplate_parent_class = g_type_class_ref (GST_TYPE_OBJECT);
gst_pad_template_signals[TEMPL_PAD_CREATED] =
g_signal_new ("pad-created", G_TYPE_FROM_CLASS (klass), G_SIGNAL_RUN_LAST,
G_STRUCT_OFFSET (GstPadTemplateClass, pad_created),
NULL, NULL, gst_marshal_VOID__OBJECT, G_TYPE_NONE, 1, GST_TYPE_PAD);
gobject_class->dispose = gst_pad_template_dispose;
gstobject_class->path_string_separator = "*";
}
static void
gst_pad_template_init (GstPadTemplate * templ)
{
}
static void
gst_pad_template_dispose (GObject * object)
{
GstPadTemplate *templ = GST_PAD_TEMPLATE (object);
g_free (GST_PAD_TEMPLATE_NAME_TEMPLATE (templ));
if (GST_PAD_TEMPLATE_CAPS (templ)) {
gst_caps_free (GST_PAD_TEMPLATE_CAPS (templ));
}
G_OBJECT_CLASS (padtemplate_parent_class)->dispose (object);
}
/* ALWAYS padtemplates cannot have conversion specifications, it doesn't make
* sense.
* SOMETIMES padtemplates can do whatever they want, they are provided by the
* element.
* REQUEST padtemplates can be reverse-parsed (the user asks for 'sink1', the
* 'sink%d' template is automatically selected), so we need to restrict their
* naming.
*/
static gboolean
name_is_valid (const gchar * name, GstPadPresence presence)
{
const gchar *str;
if (presence == GST_PAD_ALWAYS) {
if (strchr (name, '%')) {
g_warning ("invalid name template %s: conversion specifications are not"
" allowed for GST_PAD_ALWAYS padtemplates", name);
return FALSE;
}
} else if (presence == GST_PAD_REQUEST) {
if ((str = strchr (name, '%')) && strchr (str + 1, '%')) {
g_warning ("invalid name template %s: only one conversion specification"
" allowed in GST_PAD_REQUEST padtemplate", name);
return FALSE;
}
if (str && (*(str + 1) != 's' && *(str + 1) != 'd')) {
g_warning ("invalid name template %s: conversion specification must be of"
" type '%%d' or '%%s' for GST_PAD_REQUEST padtemplate", name);
return FALSE;
}
if (str && (*(str + 2) != '\0')) {
g_warning ("invalid name template %s: conversion specification must"
" appear at the end of the GST_PAD_REQUEST padtemplate name", name);
return FALSE;
}
}
return TRUE;
}
/**
* gst_static_pad_template_get:
* @pad_template: the static pad template
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Converts a #GstStaticPadTemplate into a #GstPadTemplate.
*
* Returns: a new #GstPadTemplate.
*/
GstPadTemplate *
gst_static_pad_template_get (GstStaticPadTemplate * pad_template)
{
GstPadTemplate *new;
if (!name_is_valid (pad_template->name_template, pad_template->presence))
return NULL;
new = g_object_new (gst_pad_template_get_type (),
"name", pad_template->name_template, NULL);
GST_PAD_TEMPLATE_NAME_TEMPLATE (new) = g_strdup (pad_template->name_template);
GST_PAD_TEMPLATE_DIRECTION (new) = pad_template->direction;
GST_PAD_TEMPLATE_PRESENCE (new) = pad_template->presence;
GST_PAD_TEMPLATE_CAPS (new) =
gst_caps_copy (gst_static_caps_get (&pad_template->static_caps));
return new;
}
/**
* gst_pad_template_new:
* @name_template: the name template.
* @direction: the #GstPadDirection of the template.
* @presence: the #GstPadPresence of the pad.
* @caps: a #GstCaps set for the template. The caps are taken ownership of.
*
* Creates a new pad template with a name according to the given template
* and with the given arguments. This functions takes ownership of the provided
* caps, so be sure to not use them afterwards.
*
* Returns: a new #GstPadTemplate.
*/
GstPadTemplate *
gst_pad_template_new (const gchar * name_template,
GstPadDirection direction, GstPadPresence presence, GstCaps * caps)
{
GstPadTemplate *new;
g_return_val_if_fail (name_template != NULL, NULL);
g_return_val_if_fail (caps != NULL, NULL);
g_return_val_if_fail (direction == GST_PAD_SRC
|| direction == GST_PAD_SINK, NULL);
g_return_val_if_fail (presence == GST_PAD_ALWAYS
|| presence == GST_PAD_SOMETIMES || presence == GST_PAD_REQUEST, NULL);
if (!name_is_valid (name_template, presence))
return NULL;
#if 0
#ifdef USE_POISONING
if (caps) {
GstCaps *newcaps = gst_caps_copy (caps);
gst_caps_free (caps);
caps = newcaps;
}
#endif
#endif
new = g_object_new (gst_pad_template_get_type (),
"name", name_template, NULL);
GST_PAD_TEMPLATE_NAME_TEMPLATE (new) = g_strdup (name_template);
GST_PAD_TEMPLATE_DIRECTION (new) = direction;
GST_PAD_TEMPLATE_PRESENCE (new) = presence;
GST_PAD_TEMPLATE_CAPS (new) = caps;
return new;
}
/**
* gst_pad_template_get_caps:
* @templ: a #GstPadTemplate to get capabilities of.
*
* Gets the capabilities of the pad template.
*
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
* Returns: the #GstCaps of the pad template. If you need to keep a reference to
* the caps, make a copy (see gst_caps_copy ()).
*/
const GstCaps *
gst_pad_template_get_caps (GstPadTemplate * templ)
{
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (GST_IS_PAD_TEMPLATE (templ), NULL);
return GST_PAD_TEMPLATE_CAPS (templ);
}
/**
* gst_pad_set_element_private:
* @pad: the #GstPad to set the private data of.
* @priv: The private data to attach to the pad.
*
* Set the given private data gpointer on the pad.
* This function can only be used by the element that owns the pad.
*/
void
gst_pad_set_element_private (GstPad * pad, gpointer priv)
{
pad->element_private = priv;
}
/**
* gst_pad_get_element_private:
* @pad: the #GstPad to get the private data of.
*
* Gets the private data of a pad.
*
* Returns: a #gpointer to the private data.
*/
gpointer
gst_pad_get_element_private (GstPad * pad)
{
return pad->element_private;
}
/***** ghost pads *****/
GType _gst_ghost_pad_type = 0;
static void gst_ghost_pad_class_init (GstGhostPadClass * klass);
static void gst_ghost_pad_init (GstGhostPad * pad);
static void gst_ghost_pad_dispose (GObject * object);
static void gst_ghost_pad_get_property (GObject * object, guint prop_id,
GValue * value, GParamSpec * pspec);
static void gst_ghost_pad_set_property (GObject * object, guint prop_id,
const GValue * value, GParamSpec * pspec);
static GstPad *ghost_pad_parent_class = NULL;
/* static guint gst_ghost_pad_signals[LAST_SIGNAL] = { 0 }; */
enum
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
{
GPAD_ARG_0,
GPAD_ARG_REAL_PAD
/* fill me */
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
};
GType
gst_ghost_pad_get_type (void)
{
if (!_gst_ghost_pad_type) {
static const GTypeInfo pad_info = {
sizeof (GstGhostPadClass), NULL, NULL,
(GClassInitFunc) gst_ghost_pad_class_init, NULL, NULL,
sizeof (GstGhostPad),
0,
(GInstanceInitFunc) gst_ghost_pad_init,
NULL
};
_gst_ghost_pad_type = g_type_register_static (GST_TYPE_PAD, "GstGhostPad",
&pad_info, 0);
}
return _gst_ghost_pad_type;
}
static void
gst_ghost_pad_class_init (GstGhostPadClass * klass)
{
GObjectClass *gobject_class;
gobject_class = (GObjectClass *) klass;
ghost_pad_parent_class = g_type_class_ref (GST_TYPE_PAD);
gobject_class->dispose = GST_DEBUG_FUNCPTR (gst_ghost_pad_dispose);
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
gobject_class->set_property = GST_DEBUG_FUNCPTR (gst_ghost_pad_set_property);
gobject_class->get_property = GST_DEBUG_FUNCPTR (gst_ghost_pad_get_property);
g_object_class_install_property (gobject_class, GPAD_ARG_REAL_PAD,
g_param_spec_object ("real-pad", "Real pad",
"The real pad for the ghost pad", GST_TYPE_PAD, G_PARAM_READWRITE));
}
static void
gst_ghost_pad_init (GstGhostPad * pad)
{
/* zeroed by glib */
}
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
static void
gst_ghost_pad_dispose (GObject * object)
{
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
g_object_set (object, "real-pad", NULL, NULL);
G_OBJECT_CLASS (ghost_pad_parent_class)->dispose (object);
}
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
static void
gst_ghost_pad_set_property (GObject * object, guint prop_id,
const GValue * value, GParamSpec * pspec)
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
{
GstPad *ghostpad = (GstPad *) object;
GstPad *oldrealpad = (GstPad *) GST_GPAD_REALPAD (ghostpad);
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
GstPad *realpad = NULL;
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
switch (prop_id) {
case GPAD_ARG_REAL_PAD:
realpad = g_value_get_object (value);
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
if (oldrealpad) {
if (realpad == oldrealpad)
return;
else
gst_pad_remove_ghost_pad (oldrealpad, ghostpad);
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
}
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
if (realpad)
gst_pad_add_ghost_pad (realpad, ghostpad);
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
break;
}
}
static void
gst_ghost_pad_get_property (GObject * object, guint prop_id,
GValue * value, GParamSpec * pspec)
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
{
switch (prop_id) {
case GPAD_ARG_REAL_PAD:
g_value_set_object (value, GST_GPAD_REALPAD (object));
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
break;
}
}
/**
* gst_ghost_pad_new:
* @name: the name of the new ghost pad.
* @pad: the #GstPad to create a ghost pad for.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Creates a new ghost pad associated with @pad, and named @name. If @name is
* %NULL, a guaranteed unique name (across all ghost pads) will be assigned.
*
gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. Original commit message from CVS: 2004-02-10 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (gst_pad_custom_new): Add a FIXME, this is a hacky way to do inheritance. (gst_pad_get_event_masks, gst_pad_get_event_masks_default) (gst_pad_get_query_types, gst_pad_get_query_types_default): Routine docs. (gst_pad_set_link_function, gst_pad_set_fixate_function) (gst_pad_set_getcaps_function): Doc from Dave's negotation random doc. (gst_pad_unlink, gst_pad_is_linked): Docs. (gst_pad_renegotiate): A brief description of capsnego. (gst_pad_try_set_caps): Document. (gst_pad_try_set_caps_nonfixed): Document. (gst_pad_can_link_filtered, gst_pad_link_filtered): Doc fixes. (gst_pad_set_parent): Deprecated (although not out of the API). (gst_pad_get_parent): Deprecated, although many plugins use this. (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): Doc that these are private and will go away in 0.9. (gst_pad_perform_negotiate): Doc. (gst_pad_link_unnegotiate): I think this is meant to be static. (gst_pad_get_negotiated_caps, gst_pad_get_pad_template_caps) (gst_pad_template_get_caps_by_name, gst_pad_check_compatibility) (gst_pad_get_peer): Doc updates. (gst_pad_caps_change_notify): Doc. (gst_pad_alloc_buffer, gst_pad_push, gst_static_pad_template_get) (gst_ghost_pad_new): Doc fixes. * gst/gstobject.c (gst_object_get_parent, gst_object_unparent) (gst_object_check_uniqueness): * gst/gstelement.c (gst_element_add_pad) (gst_element_add_ghost_pad, gst_element_remove_pad) (gst_element_remove_ghost_pad, gst_element_get_pad) (gst_element_get_static_pad, gst_element_get_pad_list) (gst_element_class_get_pad_template_list) (gst_element_class_get_pad_template): Work on the docs. (gst_element_get_pad_template_list): Uses the class method. (gst_element_get_compatible_pad_template): Docs, and consolidate some test conditions. (gst_element_get_pad_from_template): New static function. (gst_element_request_compatible_pad): Docs, and work with non-request compatible templates. (gst_element_get_compatible_pad_filtered): Docs and remove redundant checks. (gst_element_get_compatible_pad, gst_element_link_pads_filtered) (gst_element_link_filtered, gst_element_link_many) (gst_element_link, gst_element_link_pads) (gst_element_unlink_many): Docs. 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-11 13:26:04 +00:00
* Returns: a new ghost #GstPad, or %NULL in case of an error.
*/
GstPad *
gst_ghost_pad_new (const gchar * name, GstPad * pad)
{
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
GstPad *gpad;
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
gpad = g_object_new (GST_TYPE_GHOST_PAD, "name", name, "real-pad", pad, NULL);
GST_CAT_DEBUG (GST_CAT_PADS, "created ghost pad \"%s\" for pad %s:%s",
GST_OBJECT_NAME (gpad), GST_DEBUG_PAD_NAME (pad));
gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. Original commit message from CVS: 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstpad.c (_gst_real_pad_fixate_accumulator): s/pointer/boxed/. * gst/gstmarshal.list (VOID:BOXED, BOXED:BOXED): New marshallers. * gst/gstpad.c (gst_real_pad_class_init): Use a BOXED:BOXED marshaller for ::fixate, and VOID:BOXED for ::caps-nego-failed, with the type=GST_TYPE_CAPS. This allows language bindings to know what kind of data they're dealing with. * gst/gstcaps.c (_gst_caps_value_init): GBoxed values initialize to NULL when g_value_init is called. GstCaps, which rolls its own type implementation, now does the same instead of allocating empty caps. (_gst_caps_initialize, _gst_caps_collect_value, _gst_caps_lcopy_value): Provide collect_value and lcopy_value type table methods. This allows G_VALUE_COLLECT to work. 2004-02-05 Andy Wingo <wingo@pobox.com> * configure.ac: * testsuite/Makefile.am (SUBDIRS): * testsuite/ghostpads/Makefile.am: * testsuite/ghostpads/ghostpads.c: A new test for ghost pads. * gst/gstpad.c (gst_pad_add_ghost_pad, gst_pad_remove_ghost_pad): These two routines are the only ones that set GST_GPAD_REALPAD(gpad), the ghost pad list, and the ghost pad's pad template. They should be made static, depending on ABI needs. (gst_real_pad_dispose): Handle the case of ghost pads without a parent. Assert after dealing with ghost pads that the ghost pad list is empty. (gst_ghost_pad_class_init): New property added, ::real-pad. Can be set after creation. (gst_ghost_pad_dispose): Set ::real-pad to NULL. (gst_ghost_pad_set_property, gst_ghost_pad_get_property): New functions. set_property will call add_ghost_pad/remove_ghost_pad as appropriate. (gst_ghost_pad_new): All the work is offloaded to g_object_new. * gst/gstelement.c (gst_element_add_pad): Handle ghost pads as well. (gst_element_add_ghost_pad): Remove code duplicated from _add_pad. (gst_element_remove_pad): Handle ghost pads as well. (gst_element_remove_ghost_pad): Deprecated (could be removed, depending on API-stability needs). 2004-02-05 Andy Wingo <wingo@pobox.com> * gst/gstbin.[ch]: (gst_bin_get_by_interface): GTypes are scalars, of course they're const
2004-02-05 13:51:26 +00:00
return gpad;
}
/**
* gst_pad_get_internal_links_default:
* @pad: the #GstPad to get the internal links of.
*
* Gets a list of pads to which the given pad is linked to
* inside of the parent element.
* This is the default handler, and thus returns a list of all of the
* pads inside the parent element with opposite direction.
* The caller must free this list after use.
*
* Returns: a newly allocated #GList of pads.
*/
GList *
gst_pad_get_internal_links_default (GstPad * pad)
{
GList *res = NULL;
GstElement *parent;
GList *parent_pads;
GstPadDirection direction;
GstRealPad *rpad;
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
rpad = GST_PAD_REALIZE (pad);
direction = rpad->direction;
parent = GST_PAD_PARENT (rpad);
parent_pads = parent->pads;
while (parent_pads) {
GstRealPad *parent_pad = GST_PAD_REALIZE (parent_pads->data);
if (parent_pad->direction != direction) {
res = g_list_prepend (res, parent_pad);
}
parent_pads = g_list_next (parent_pads);
}
return res;
}
/**
* gst_pad_get_internal_links:
* @pad: the #GstPad to get the internal links of.
*
* Gets a list of pads to which the given pad is linked to
* inside of the parent element.
* The caller must free this list after use.
*
* Returns: a newly allocated #GList of pads.
*/
GList *
gst_pad_get_internal_links (GstPad * pad)
{
GList *res = NULL;
GstRealPad *rpad;
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
rpad = GST_PAD_REALIZE (pad);
if (GST_RPAD_INTLINKFUNC (rpad))
res = GST_RPAD_INTLINKFUNC (rpad) (GST_PAD (rpad));
return res;
}
static gboolean
gst_pad_event_default_dispatch (GstPad * pad, GstElement * element,
GstEvent * event)
{
GList *orig, *pads;
GST_INFO_OBJECT (pad, "Sending event %p to all internally linked pads",
event);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
orig = pads = gst_pad_get_internal_links (pad);
while (pads) {
GstPad *eventpad = GST_PAD (pads->data);
pads = g_list_next (pads);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
/* for all of the internally-linked pads that are actually linked */
if (GST_PAD_IS_LINKED (eventpad)) {
if (GST_PAD_DIRECTION (eventpad) == GST_PAD_SRC) {
/* for each pad we send to, we should ref the event; it's up
* to downstream to unref again when handled. */
GST_LOG_OBJECT (pad, "Reffing and sending event %p to %s:%s", event,
GST_DEBUG_PAD_NAME (eventpad));
gst_event_ref (event);
gst_pad_push (eventpad, GST_DATA (event));
} else {
GstPad *peerpad = GST_PAD (GST_RPAD_PEER (eventpad));
/* we only send the event on one pad, multi-sinkpad elements
* should implement a handler */
g_list_free (orig);
GST_LOG_OBJECT (pad, "sending event %p to one sink pad %s:%s", event,
GST_DEBUG_PAD_NAME (eventpad));
return gst_pad_send_event (peerpad, event);
}
}
}
/* we handled the incoming event so we unref once */
GST_LOG_OBJECT (pad, "handled event %p, unreffing", event);
gst_event_unref (event);
g_list_free (orig);
return (GST_PAD_DIRECTION (pad) == GST_PAD_SINK);
}
/**
* gst_pad_event_default:
* @pad: a #GstPad to call the default event handler on.
* @event: the #GstEvent to handle.
*
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
* Invokes the default event handler for the given pad. End-of-stream and
* discontinuity events are handled specially, and then the event is sent to all
* pads internally linked to @pad. Note that if there are many possible sink
* pads that are internally linked to @pad, only one will be sent an event.
* Multi-sinkpad elements should implement custom event handlers.
*
* Returns: TRUE if the event was sent succesfully.
*/
gboolean
gst_pad_event_default (GstPad * pad, GstEvent * event)
{
GstElement *element;
g_return_val_if_fail (GST_IS_PAD (pad), FALSE);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (event != NULL, FALSE);
element = GST_PAD_PARENT (pad);
switch (GST_EVENT_TYPE (event)) {
case GST_EVENT_EOS:
gst_pad_event_default_dispatch (pad, element, event);
gst_element_set_eos (element);
break;
case GST_EVENT_DISCONTINUOUS:
{
gint64 time;
if (gst_element_requires_clock (element) && element->clock) {
if (gst_event_discont_get_value (event, GST_FORMAT_TIME, &time)) {
gst_element_set_time (element, time);
} else {
GstFormat format = GST_FORMAT_TIME;
guint i;
for (i = 0; i < event->event_data.discont.noffsets; i++) {
if (gst_pad_convert (pad,
event->event_data.discont.offsets[i].format,
event->event_data.discont.offsets[i].value, &format,
&time)) {
gst_element_set_time (element, time);
} else if (i == event->event_data.discont.noffsets) {
g_warning
("can't adjust clock to new time when time not provided");
}
}
}
}
}
default:
return gst_pad_event_default_dispatch (pad, element, event);
}
return TRUE;
}
/**
* gst_pad_dispatcher:
* @pad: a #GstPad to dispatch.
* @dispatch: the #GstDispatcherFunction to call.
* @data: gpointer user data passed to the dispatcher function.
*
* Invokes the given dispatcher function on all pads that are
* internally linked to the given pad.
* The GstPadDispatcherFunction should return TRUE when no further pads
* need to be processed.
*
* Returns: TRUE if one of the dispatcher functions returned TRUE.
*/
gboolean
gst_pad_dispatcher (GstPad * pad, GstPadDispatcherFunction dispatch,
gpointer data)
{
gboolean res = FALSE;
GList *int_pads, *orig;
g_return_val_if_fail (GST_IS_PAD (pad), FALSE);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (dispatch != NULL, FALSE);
orig = int_pads = gst_pad_get_internal_links (pad);
while (int_pads) {
GstRealPad *int_rpad = GST_PAD_REALIZE (int_pads->data);
GstRealPad *int_peer = GST_RPAD_PEER (int_rpad);
if (int_peer) {
res = dispatch (GST_PAD (int_peer), data);
if (res)
break;
}
int_pads = g_list_next (int_pads);
}
g_list_free (orig);
return res;
}
/**
* gst_pad_send_event:
* @pad: a #GstPad to send the event to.
* @event: the #GstEvent to send to the pad.
*
* Sends the event to the pad.
*
* Returns: TRUE if the event was handled.
*/
gboolean
gst_pad_send_event (GstPad * pad, GstEvent * event)
{
gboolean success = FALSE;
GstRealPad *rpad;
GstElement *parent;
g_return_val_if_fail (GST_IS_PAD (pad), FALSE);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (event != NULL, FALSE);
parent = gst_pad_get_parent (pad);
g_return_val_if_fail (GST_STATE (parent) >= GST_STATE_PAUSED, FALSE);
rpad = GST_PAD_REALIZE (pad);
if (GST_EVENT_SRC (event) == NULL)
GST_EVENT_SRC (event) = gst_object_ref (GST_OBJECT (rpad));
GST_CAT_DEBUG (GST_CAT_EVENT, "have event type %d on pad %s:%s",
GST_EVENT_TYPE (event), GST_DEBUG_PAD_NAME (rpad));
if (GST_RPAD_EVENTHANDLER (rpad))
success = GST_RPAD_EVENTHANDLER (rpad) (GST_PAD (rpad), event);
else {
g_warning ("pad %s:%s has no event handler", GST_DEBUG_PAD_NAME (rpad));
gst_event_unref (event);
}
return success;
}
typedef struct
{
GstFormat src_format;
gint64 src_value;
GstFormat *dest_format;
gint64 *dest_value;
}
GstPadConvertData;
static gboolean
gst_pad_convert_dispatcher (GstPad * pad, GstPadConvertData * data)
{
return gst_pad_convert (pad, data->src_format, data->src_value,
data->dest_format, data->dest_value);
}
/**
* gst_pad_convert_default:
* @pad: a #GstPad to invoke the default converter on.
* @src_format: the source #GstFormat.
* @src_value: the source value.
* @dest_format: a pointer to the destination #GstFormat.
* @dest_value: a pointer to the destination value.
*
* Invokes the default converter on a pad.
* This will forward the call to the pad obtained
* using the internal link of
* the element.
*
* Returns: TRUE if the conversion could be performed.
*/
gboolean
gst_pad_convert_default (GstPad * pad,
GstFormat src_format, gint64 src_value,
GstFormat * dest_format, gint64 * dest_value)
{
GstPadConvertData data;
g_return_val_if_fail (GST_IS_PAD (pad), FALSE);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (dest_format != NULL, FALSE);
g_return_val_if_fail (dest_value != NULL, FALSE);
data.src_format = src_format;
data.src_value = src_value;
data.dest_format = dest_format;
data.dest_value = dest_value;
return gst_pad_dispatcher (pad, (GstPadDispatcherFunction)
gst_pad_convert_dispatcher, &data);
}
/**
* gst_pad_convert:
* @pad: a #GstPad to invoke the default converter on.
* @src_format: the source #GstFormat.
* @src_value: the source value.
* @dest_format: a pointer to the destination #GstFormat.
* @dest_value: a pointer to the destination value.
*
* Invokes a conversion on the pad.
*
* Returns: TRUE if the conversion could be performed.
*/
gboolean
gst_pad_convert (GstPad * pad,
GstFormat src_format, gint64 src_value,
GstFormat * dest_format, gint64 * dest_value)
{
GstRealPad *rpad;
g_return_val_if_fail (GST_IS_PAD (pad), FALSE);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (dest_format != NULL, FALSE);
g_return_val_if_fail (dest_value != NULL, FALSE);
if (src_format == *dest_format) {
*dest_value = src_value;
return TRUE;
}
rpad = GST_PAD_REALIZE (pad);
if (GST_RPAD_CONVERTFUNC (rpad)) {
return GST_RPAD_CONVERTFUNC (rpad) (GST_PAD (rpad), src_format,
src_value, dest_format, dest_value);
}
return FALSE;
}
typedef struct
{
GstQueryType type;
GstFormat *format;
gint64 *value;
}
GstPadQueryData;
static gboolean
gst_pad_query_dispatcher (GstPad * pad, GstPadQueryData * data)
{
return gst_pad_query (pad, data->type, data->format, data->value);
}
/**
* gst_pad_query_default:
* @pad: a #GstPad to invoke the default query on.
* @type: the #GstQueryType of the query to perform.
* @format: a pointer to the #GstFormat of the result.
* @value: a pointer to the result.
*
* Invokes the default query function on a pad.
*
* Returns: TRUE if the query could be performed.
*/
gboolean
gst_pad_query_default (GstPad * pad, GstQueryType type,
GstFormat * format, gint64 * value)
{
GstPadQueryData data;
g_return_val_if_fail (GST_IS_PAD (pad), FALSE);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (format != NULL, FALSE);
g_return_val_if_fail (value != NULL, FALSE);
data.type = type;
data.format = format;
data.value = value;
return gst_pad_dispatcher (pad, (GstPadDispatcherFunction)
gst_pad_query_dispatcher, &data);
}
/**
* gst_pad_query:
* @pad: a #GstPad to invoke the default query on.
* @type: the #GstQueryType of the query to perform.
* @format: a pointer to the #GstFormat asked for.
* On return contains the #GstFormat used.
* @value: a pointer to the result.
*
* Queries a pad for one of the available properties. The format will be
* adjusted to the actual format used when specifying formats such as
* GST_FORMAT_DEFAULT.
* FIXME: Tell if the format can be adjusted when specifying a definite format.
*
* Returns: TRUE if the query could be performed.
*/
gboolean
gst_pad_query (GstPad * pad, GstQueryType type,
GstFormat * format, gint64 * value)
{
GstRealPad *rpad;
g_return_val_if_fail (GST_IS_PAD (pad), FALSE);
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (format != NULL, FALSE);
g_return_val_if_fail (value != NULL, FALSE);
rpad = GST_PAD_REALIZE (pad);
g_return_val_if_fail (rpad, FALSE);
if (GST_RPAD_QUERYFUNC (rpad))
return GST_RPAD_QUERYFUNC (rpad) (GST_PAD (rpad), type, format, value);
return FALSE;
}
static gboolean
gst_pad_get_formats_dispatcher (GstPad * pad, const GstFormat ** data)
{
*data = gst_pad_get_formats (pad);
return TRUE;
}
/**
* gst_pad_get_formats_default:
* @pad: a #GstPad to query
*
* Invoke the default format dispatcher for the pad.
*
* Returns: An array of GstFormats ended with a 0 value.
*/
const GstFormat *
gst_pad_get_formats_default (GstPad * pad)
{
GstFormat *result = NULL;
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
gst_pad_dispatcher (pad, (GstPadDispatcherFunction)
gst_pad_get_formats_dispatcher, &result);
return result;
}
/**
* gst_pad_get_formats:
* @pad: a #GstPad to query
*
* Gets the list of supported formats from the pad.
*
* Returns: An array of GstFormats ended with a 0 value.
*/
const GstFormat *
gst_pad_get_formats (GstPad * pad)
{
GstRealPad *rpad;
gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. Original commit message from CVS: 2004-02-24 Andy Wingo <wingo@pobox.com> * gst/gstelement.c (gst_element_dispose): Protect against multiple invocations. * gst/schedulers/gstoptimalscheduler.c I added a mess of prototypes at the top of the file by way of documentation. Some of the operations on chains and groups were re-organized. (create_group): Added a type argument so if the group is enabled, the setup_group_scheduler knows what to do. (group_elements): Added a type argument here, too, to be passed on to create_group. (group_element_set_enabled): If an unlinked PLAYING element is added to a bin, we have to create a new group to hold the element, and this function will be called before the group is added to the chain. Thus we have a valid case for group->chain==NULL. Instead of calling chain_group_set_enabled, just set the flag on the group (the chain's status will be set when the group is added to it). (gst_opt_scheduler_state_transition, chain_group_set_enabled): Setup the group scheduler when the group is enabled, not specifically when an element goes PAUSED->PLAYING. This means PLAYING elements can be added, linked, and scheduled into a PLAYING pipeline, as was intended. (add_to_group): Don't ref the group twice. I don't know when this double-ref got in here. Removing it has the potential to cause segfaults if other parts of the scheduler are buggy. If you find that the scheduler is segfaulting for you, put in an extra ref here and see if that hacks over the underlying issue. Of course, then find out what code is unreffing a group it doesn't own... (create_group): Make the extra refcount floating, and remove it after adding the element. This means that... (unref_group): Destroy when the refcount reaches 0, not 1, like every other refcounted object in the known universe. (remove_from_group): When a group becomes empty, set it to be not active, and remove it from its chain. Don't unref it again, there's no floating reference any more. (destroy_group): We have to remove the group from the chain in remove_from_group (rather than here) to break refcounting cycles (the chain always has a ref on the group). So assert that group->chain==NULL. (ref_group_by_count): Removed, it was commented out anyway. (merge_chains): Use the remove_from_chain and add_to_chain primitives to do the reparenting, instead of rolling our own implementation. (add_to_chain): The first non-disabled group in the chain's group list will be the entry point for the chain. Because buffers can accumulate in loop elements' peer bufpens, we preferentially schedule loop groups before get groups to avoid unnecessary execution of get-based groups when the bufpens are already full. (gst_opt_scheduler_schedule_run_queue): Debug fixes. (get_group_schedule_function): Ditto. (loop_group_schedule_function): Ditto. (gst_opt_scheduler_loop_wrapper): Ditto. (gst_opt_scheduler_iterate): Ditto. I understand the opt scheduler now, yippee! * gst/gstpad.c: All throughout, added FIXMEs to look at for 0.9. (gst_pad_get_name, gst_pad_set_chain_function) (gst_pad_set_get_function, gst_pad_set_event_function) (gst_pad_set_event_mask_function, gst_pad_get_event_masks) (gst_pad_get_event_masks_default, gst_pad_set_convert_function) (gst_pad_set_query_function, gst_pad_get_query_types) (gst_pad_get_query_types_default) (gst_pad_set_internal_link_function) (gst_pad_set_formats_function, gst_pad_set_link_function) (gst_pad_set_fixate_function, gst_pad_set_getcaps_function) (gst_pad_set_bufferalloc_function, gst_pad_unlink) (gst_pad_renegotiate, gst_pad_set_parent, gst_pad_get_parent) (gst_pad_add_ghost_pad, gst_pad_proxy_getcaps) (gst_pad_proxy_pad_link, gst_pad_proxy_fixate) (gst_pad_get_pad_template_caps, gst_pad_check_compatibility) (gst_pad_get_peer, gst_pad_get_allowed_caps) (gst_pad_alloc_buffer, gst_pad_push, gst_pad_pull) (gst_pad_selectv, gst_pad_select, gst_pad_template_get_caps) (gst_pad_event_default_dispatch, gst_pad_event_default) (gst_pad_dispatcher, gst_pad_send_event, gst_pad_convert_default) (gst_pad_convert, gst_pad_query_default, gst_pad_query) (gst_pad_get_formats_default, gst_pad_get_formats): Better argument checks, and some doc fixes. (gst_pad_custom_new_from_template): Um, does anyone use these functions? Actually make a custom pad instead of a normal one. (gst_pad_try_set_caps): Transpose some checks. (gst_pad_try_set_caps_nonfixed): Same, and use a macro to check if the pad is in negotiation. (gst_pad_try_relink_filtered): Use pad_link_prepare. * gst/gstelement.c: Remove prototypes also defined in gstclock.h. * gst/gstelement.h: * gst/gstclock.h: Un-deprecate the old clocking API, as discussed on the list.
2004-02-25 13:16:12 +00:00
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
rpad = GST_PAD_REALIZE (pad);
if (GST_RPAD_FORMATSFUNC (rpad))
return GST_RPAD_FORMATSFUNC (rpad) (GST_PAD (pad));
return NULL;
}
#define CALL_CHAINFUNC(pad, data) G_STMT_START {\
GstData *__temp = (data); \
DEBUG_DATA (pad, __temp, "calling chain function with "); \
if (GST_IS_EVENT (__temp) && \
!GST_FLAG_IS_SET (gst_pad_get_parent (pad), GST_ELEMENT_EVENT_AWARE)) { \
gst_pad_send_event (pad, GST_EVENT (__temp)); \
} else { \
GST_FLAG_SET (pad, GST_RPAD_IN_CHAINFUNC); \
GST_RPAD_CHAINFUNC (pad) (pad, __temp); \
GST_FLAG_UNSET (pad, GST_RPAD_IN_CHAINFUNC); \
} \
}G_STMT_END
/**
* gst_pad_call_chain_function:
* @pad: sink pad to call chain function on
* @data: data to call the chain function with
*
* Calls the chain function of the given pad while making sure the internal
* consistency is kept. Use this function inside schedulers instead of calling
* the chain function yourself.
*/
void
gst_pad_call_chain_function (GstPad * pad, GstData * data)
{
GstPadLink *link;
g_return_if_fail (GST_IS_REAL_PAD (pad));
g_return_if_fail (GST_PAD_IS_SINK (pad));
g_return_if_fail (data != NULL);
g_return_if_fail (GST_RPAD_CHAINFUNC (pad) != NULL);
g_return_if_fail (GST_RPAD_LINK (pad) != NULL);
link = GST_RPAD_LINK (pad);
if (!link->engaged) {
g_assert (link->temp_store == NULL);
if (GST_IS_BUFFER (data)) {
GST_DEBUG ("moving data buffer %p back to temp_store", data);
link->temp_store = data;
link->engaged = TRUE;
CALL_CHAINFUNC (pad, _invent_event (pad, GST_BUFFER (data)));
link = GST_RPAD_LINK (pad);
if (link->temp_store == NULL) /* happens after relinking in chainfunc */
return;
g_assert (link->temp_store == data);
link->temp_store = NULL;
} else if (GST_IS_EVENT (data) &&
GST_EVENT_TYPE (data) == GST_EVENT_DISCONTINUOUS &&
GST_EVENT_DISCONT_NEW_MEDIA (data)) {
link->engaged = TRUE;
GST_CAT_LOG (GST_CAT_SCHEDULING,
"link engaged by discont event %p for pad %s:%s", data,
GST_DEBUG_PAD_NAME (pad));
}
}
CALL_CHAINFUNC (pad, data);
}
/**
* gst_pad_call_get_function:
* @pad: sink pad to call chain function on
*
* Calls the get function of the given pad while making sure the internal
* consistency is kept. Use this function inside schedulers instead of calling
* the get function yourself.
*
* Returns: the data provided by the pad or NULL if no data was available.
*/
GstData *
gst_pad_call_get_function (GstPad * pad)
{
GstData *data;
g_return_val_if_fail (GST_IS_REAL_PAD (pad), NULL);
g_return_val_if_fail (GST_PAD_IS_SRC (pad), NULL);
g_return_val_if_fail (GST_RPAD_GETFUNC (pad) != NULL, NULL);
GST_FLAG_SET (pad, GST_RPAD_IN_GETFUNC);
data = GST_RPAD_GETFUNC (pad) (pad);
GST_FLAG_UNSET (pad, GST_RPAD_IN_GETFUNC);
DEBUG_DATA (pad, data, "getfunction returned");
return data;
}