Original commit message from CVS:
* gst/schedulers/entryscheduler.c:
completely revamp the inner workings, so it's a lot easier to
understand and extend
Original commit message from CVS:
2004-04-03 Andy Wingo <wingo@pobox.com>
* gst/schedulers/gstoptimalscheduler.c (GstOptSchedulerGroupLink):
Rename ->group1 and ->group2 to ->src and ->sink, respectively.
This allows better introspection of pipeline topology.
(add_to_chain): Don't do trickery to put loop elements first;
rather, queue a chain sort by marking the chain as dirty.
(remove_from_chain): Mark the chain dirty.
(sort_chain): New function. Sorts the group list so that terminal
sinks are first. This means elements on the sink side will be
preferentially sscheduled before elements on the src side of the
pipeline.
(chain_recursively_migrate_group): Use OTHER_GROUP_LINK.
(schedule_chain): If the chain is marked DIRTY, call sort_chain.
(gst_opt_scheduler_pad_link, gst_opt_scheduler_pad_unlink)
(group_inc_link): Change argument and variable names to match the
new link structure member names (src and sink).
(group_dec_link): Add some description
Original commit message from CVS:
* gst/elements/gstbufferstore.c: (gst_buffer_store_get_buffer):
set the offset of the buffer to the requested offset
* gst/elements/gsttypefind.c: (stop_typefinding):
revert patch 1.18 (which I unfortunately don't know the reason for).
This is needed to allow downstream elements to seek. Otherwise
typefind might overwrite a previous seek by downstream elements.
This lead to errors with id3tag and typefind on some mp3s.
* gst/schedulers/entryscheduler.c: (COTHREADS_TYPE),
(gst_entry_scheduler_iterate):
be more verbose when debugging
Original commit message from CVS:
* gst/gstpad.c: (gst_pad_set_explicit_caps):
print caps in debugging output when setting caps failed
* gst/schedulers/entryscheduler.c: (COTHREADS_TYPE),
(schedule_next_element), (get_buffer), (run_chainhandler),
(element_may_start), (gst_entry_scheduler_chain_handler),
(gst_entry_scheduler_get_handler),
(gst_entry_scheduler_state_transition),
(gst_entry_scheduler_pad_link):
make this scheduler a testcase for mandatory
discont-before-first-buffer which is needed if we want to allow apps
to release the sound device.
add SCHED_ASSERT macro to print scheduler state before an assertion
triggers.
Original commit message from CVS:
* gst/gstbin.c: (gst_bin_set_element_sched),
(gst_bin_unset_element_sched):
don't add decoupled elements to schedulers - otherwise it's
impossible to control if a link to a decoupled element was already
removed from a scheduler or not.
* gst/schedulers/cothreads_compat.h:
* gst/schedulers/gthread-cothreads.h:
add COTHREADS_TYPE macro. Make do_cothread_set_func a macro so there
is no "unused" warning.
* gst/schedulers/Makefile.am:
* gst/schedulers/entryscheduler.c:
add new scheduler, based on ideas from talking to David and Martin.
It's supposed to be small and correct. Currently it's also slow (but
it's not noticable)
* examples/retag/retag.c: (main):
* testsuite/bytestream/test1.c: (main):
fix missing NULLs at end of variadic functions
* testsuite/elements/.cvsignore:
update
Original commit message from CVS:
* gst/schedulers/gstbasicscheduler.c:
(gst_basic_scheduler_src_wrapper):
* gst/schedulers/gstoptimalscheduler.c:
(gst_opt_scheduler_loop_wrapper), (gst_opt_scheduler_get_wrapper),
(pad_clear_queued), (gst_opt_scheduler_add_element),
(gst_opt_scheduler_remove_element):
fix GStreamer to not have issues with lvalue casts anymore (fixes
some bug, but bugzilla is down)
Original commit message from CVS:
* gst/gstelement.c:
add documentation about a gobject quirk where the object hasn't the
correct class pointer set on initialization
* gst/schedulers/gstbasicscheduler.c:
(gst_basic_scheduler_src_wrapper):
make sure to not run into an infinite loop
Original commit message from CVS:
* gst/schedulers/gstoptimalscheduler.c
(gst_opt_scheduler_pad_unlink): Fix bug that causes totem to
crash on MPEG playback. My boolean arithmetic is a bit rusty.
Original commit message from CVS:
* gst/schedulers/gstoptimalscheduler.c
(gst_opt_scheduler_pad_unlink): Implement unlink for elements with
links to elements within the same group, so we can finally remove
that annoying warning. Refactor the code a little bit
(group_dec_links_for_element): Split out
Original commit message from CVS:
2004-03-07 Andy Wingo <wingo@pobox.com>
* gst/gstbin.c (gst_bin_remove): Debugging fixes.
* gst/schedulers/gstoptimalscheduler.c (destroy_group): Assert
there are no links to other groups when a group is destroyed.
(gst_opt_scheduler_pad_unlink): If the unlink means an element is
removed from a group, make sure the link count to elements linked
to other pads is appropriately decremented. This really fixes
#135672.
The 1.60->1.61 patch has been reapplied in light of this fix.
* gst/gstelement.c (gst_element_dispose): Really protect against
multiple invocations this time.
Original commit message from CVS:
2004-03-03 Benjamin Otte <otte@gnome.org>
* gst/schedulers/gstoptimalscheduler.c:
revert last commit by Andy Wingo. It causes segfaults on unreffing
in Rhythmbox. (see bug #135672)
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.
Original commit message from CVS:
2004-02-20 Andy Wingo <wingo@pobox.com>
* gst/gstbin.c:
* gst/gstbuffer.c:
* gst/gstplugin.c:
* gst/registries/gstxmlregistry.c:
* gst/schedulers/gstoptimalscheduler.c: Debugging tweaks.
* gst/gstelement.c (gst_element_set_scheduler): Debugging fixes.
(gst_element_add_pad): DEBUG->INFO, some fixes.
(gst_element_get_compatible_pad_template): Just see if the
templates' caps intersect, not if one is a strict subset of the
other. This conforms more to what gst_pad_link_intersect() does.
(gst_element_class_add_pad_template): Don't memcpy the pad
template, just ref it.
(gst_element_get_compatible_pad_filtered): Clean up debug messages
* gst/gstpad.c (gst_pad_can_link_filtered): Debug a true result.
(gst_pad_link_filtered): Debug changes.
(gst_pad_link_prepare): New function, consolidated from
can_link_filtered and link_filtered.
* gst/parse/grammar.y (gst_parse_perform_link): Made INFO output
look more like that of the functions in gstelement.c
* gst/gstinfo.c (gst_debug_print_object): Put a space before the
object, and return the empty string if object is NULL.
* gst/parse/parse.l: Remove trailing newlines when calling PRINT.
* gst/parse/grammar.y (YYFPRINTF): Log bison debugging info via
LOG, not DEBUG. We still get flex info on debug.
* gst/registries/gstxmlregistry.c (gst_xml_registry_load): Make
debug string more verbose.
(plugin_times_older_than): DEBUG->LOG.
Original commit message from CVS:
2004-01-13 Benjamin Otte <in7y118@public.uni-hamburg.de>
* gst/gstclock.c: (gst_clock_class_init), (gst_clock_init),
(gst_clock_set_speed), (gst_clock_set_active),
(gst_clock_is_active), (gst_clock_reset),
(gst_clock_handle_discont):
* gst/gstclock.h:
deprecate old interface and disable functions that aren't in use
anymore.
* gst/gstelement.h:
* gst/gstelement.c: (gst_element_get_time), (gst_element_wait),
(gst_element_set_time), (gst_element_adjust_time):
add concept of "element time" and functions to get/set this time.
* gst/gstelement.c: (gst_element_change_state):
update element time correctly.
* gst/gstelement.c: (gst_element_get_compatible_pad_filtered):
This is a debug message, not a g_critical.
* gst/gstpad.c: (gst_pad_event_default):
handle discontinuous events right with element time.
* gst/gstscheduler.c: (gst_scheduler_state_transition):
update to clocking fixes.
set clocks on elements in READY=>PAUSED. The old behaviour caused
a wrong element time on the first element that started playing.
* gst/schedulers/gstbasicscheduler.c:
(gst_basic_scheduler_class_init):
* gst/schedulers/gstoptimalscheduler.c:
(gst_opt_scheduler_class_init):
remove code that just implements the default behaviour.
* gst/elements/gstfakesink.c: (gst_fakesink_chain):
update to use new clocking functions
* testsuite/clock/clock1.c: (gst_clock_debug), (main):
* testsuite/clock/clock2.c: (gst_clock_debug), (main):
update to test new element time.
* gst/autoplug/gstspideridentity.c: (gst_spider_identity_getcaps):
use _get_allowed_caps instead of _get_caps. This catches filtered
caps correctly.
* testsuite/debug/commandline.c:
update for new GST_DEBUG syntax.
* testsuite/threads/Makefile.am:
disable a test that only works sometimes.
Original commit message from CVS:
s/GstBuffer/GstData/ in the API where you can pass events. Fix the plugins to deal with that. Fixes#113488. Also includes scheduler patches, and probably fixes some queue bugs relating to events and buffers.
Original commit message from CVS:
When scheduling a group, check that its not the group thats currently in the
top of the runqueue.
Then, when scheduling a different group, disable the first group and search for the next enabled group to run. Stops deadlocking in loop based elements.
Also remove the g_list_length's from the debugging statements for a bit of speed
Original commit message from CVS:
implement translatable error messages using gerror.
Includes bugfixes for:
- crash when unlinking Ghostpads
- make *_PAD_* macros use glib casts
- make spider typefinding merge buffers correctly
Original commit message from CVS:
Removed building of ext/cothreads and corresponding scheduler. If
you want to reenable it, please fix it first. (See bug #87163)
Original commit message from CVS:
GST_DEBUG reorganization
This is a big diff (ca 450k), containing loads of stuff:
- gstinfo.[ch] complete rewrite
- changing of all GST_DEBUG messages to reflect that change
- reorganization of subsystem disabling
- addition of gstconfig.h.in so we can track the disablings
- <gst/gst.h> does not include <unistd.h> and <config.h> anymore
- documentation updated for gstinfo stuff (build the docs yourself to know what changed)
- bugfixes for making of the docs (files from CVS are not deleted anymore
- testsuite for debugging changes in testsuite/debug
expect breakage
Original commit message from CVS:
Change GST_DEBUG_FUNCPTR() to return a pointer of the same type that it
is called with. Fix lots of warnings due to change.
Original commit message from CVS:
- fix unlink between decoupled elements
- clear the entry point of a group when we remove the element
- fix a potential problem in _merge_groups that could be triggered when
someone is holding an extra refcount to the group.
Original commit message from CVS:
added (hopefully) compatible cothreads emulation by using GThreads.
use '--gst-scheduler=basicgthread' to try it out
includes bugfix for opt to call do_cothreads_init when using threads
Original commit message from CVS:
- More refcounting fixes that now sustain abuse from spider, key is to be
careful when iterating over a list when our current pointer could be
removed.
Original commit message from CVS:
Fix various inconsistencies discovered while attempting to fix --disable-*.
Uraeus: this should fix the bison.simple problem that you've had.
Original commit message from CVS:
- Use the eventhandler instead of the event function to send events.
- make the scheduler setup the eventhandlers
- intercept flush events on pads links that can potentially queue data
and flush it.
- some more debugging info in spider.
I can think of one case where this flush might fail: unconnected pads where
the scheduler has not set up the eventhandler yet. I'll come up with a
solution for that soon.
Original commit message from CVS:
- added 'remove' argument to _chain_recursive_add to remove elements from
old chain before adding to new one
- reworked _pad_unlink to be simpler and more correct
Original commit message from CVS:
- some minor whitespace cleanups
- rework _chain_add_element to put the element in either the main or
disabled lists depending on whether the element is PLAYING or not.
Original commit message from CVS:
Provide intrastructure to not have to pass NULL buffers on errors and
interrupts, this should fix some issues with the optimal scheduler.
Original commit message from CVS:
- Remove annoying \n from cothread INFO/DEBUG messages
- only set an element to PAUSED when it gave an error during PLAYING
- ensure internal consistency of opt scheduler structures
- Added mex_recursion to opt scheduler
Original commit message from CVS:
- disable pads when going to PAUSED, we want to make sure no data is
passing when an element is not PLAYING.
- changed the clock sync API, element should now get a ClockID first and
sync on that. This makes it possible to cancel clock requests.
Original commit message from CVS:
- Cleanups
- Added padding to structs
- fixed typechecking/casts
- reduced casts
- implemented remove_element in gstbin
- implemented set index on bin
Original commit message from CVS:
- Factor out some error functions
- handle the cases where multipad get to multi-sink loop/chain based
elements are connected.
Original commit message from CVS:
- Removed old deprecated fastscheduler
- ifdef out cothread specific code in optimalscheduler
- added more g_asserts to optimalscheduler
- create separate scheduler called "opt", removed property from scheduler
- fixed iterations property
Original commit message from CVS:
- Reworked the clock to prepare for async notifications
- moved some common scheduler checking to gstbin
- added some vmethods to gstbin for future use
- more fixes to the optimal scheduler
- use new clock api in the schedulers
