Original commit message from CVS:
First THREADED backport attempt, focusing on adding locks and
making sure the API is threadsafe. Needs more work. More docs
follow this week.
Original commit message from CVS:
* docs/pwg/advanced-types.xml:
Fix description for buffer-frames=0.
* docs/gst/tmpl/gstbin.sgml:
* gst/gstbin.c: (gst_bin_child_state_change_func),
(gst_bin_change_state), (gst_bin_change_state_norecurse):
* gst/gstbin.h:
* testsuite/threads/Makefile.am:
* testsuite/threads/threadi.c: (cb_timeout), (cb_quit), (cb_eos),
(cb_state), (cb_play), (main):
Fix non-recursive state changes to *really* change the state
of the object, and not just call parent_class->state_change.
Fix a lot of lockups caused by this. Fixes#132775. Add test
for the problem. Also enable test to show #142588 (fixed).
* gst/gstthread.c: (gst_thread_change_state),
(gst_thread_child_state_change):
Don't exit the thread if we go to NULL and are inside thread
context. Instead, return control to the main thread context
and exit from there.
* gst/gstelement.c: (gst_element_disable_threadsafe_properties):
Don't unset virtual functions, since those may still be used.
That's not necessarily correct, but suffices for now.
* configure.ac:
* testsuite/Makefile.am:
* testsuite/pad/Makefile.am:
* testsuite/pad/chainnopull.c: (gst_test_sink_class_init),
(gst_test_sink_base_init), (gst_test_sink_chain),
(gst_test_sink_init), (main):
* testsuite/pad/getnopush.c: (gst_test_src_class_init),
(gst_test_src_base_init), (gst_test_src_get), (gst_test_src_init),
(main):
* testsuite/pad/link.c: (gst_test_element_class_init),
(gst_test_element_base_init), (gst_test_src_get),
(gst_test_src_loop), (gst_test_src_init), (gst_test_filter_chain),
(gst_test_filter_loop), (gst_test_filter_init),
(gst_test_sink_chain), (gst_test_sink_loop), (gst_test_sink_init),
(cb_error), (main):
Add tests to show #150546. Pass, but should fail (currently
disabled from the testsuite).
* gst/gstscheduler.c: (gst_scheduler_dispose):
Dereference child schedulers on dispose (#94464).
* testsuite/bytestream/filepadsink.c: (gst_fp_sink_init):
Fix typo.
* testsuite/threads/thread.c: (main):
Add more debug.
Original commit message from CVS:
* gst/gstbin.c: (gst_bin_get_type), (gst_bin_class_init),
(gst_bin_fix_state), (gst_bin_add_func), (gst_bin_remove_func),
(gst_bin_child_state_change_func), (set_kid_state_func),
(gst_bin_set_state), (gst_bin_change_state_norecurse):
make state changes work correctly and reentrant (so removing
elements from bins during state changes of bins doesn't cause
segfaults or even wrong states)
add debugging category and debugging output to print children states
* gst/gstbin.c: (gst_bin_dispose):
add some assertion checks
* gst/gstbin.h:
* gst/gstbin.c: (gst_bin_sync_children_state):
deprecate this function - it just does gst_bin_set_state (bin,
GST_STATE (bin))
* testsuite/threads/queue.c: (main):
don't use gst_bin_sync_children_state anymore
* testsuite/states/Makefile.am:
* testsuite/states/bin.c:
test that the state changes of bins work as expected
* gst/gstthread.c: (gst_thread_class_init), (gst_thread_set_state):
some adjustments to change states correctly, too
* gst/gstthread.c: (gst_thread_change_state):
don't enable/disable "threadsafe" properties, they're unused and
cause random segfaults
* testsuite/threads/Makefile.am:
the queue check randomly passes now, ignore it
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
Original commit message from CVS:
- make child_state_change a virtual function
- remove unneeded != NULL checks
- implement gst_bin_sync_children_state
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:
- 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
Original commit message from CVS:
- Removed unused locking from the cothreads
- use G_*_DECLS in .h files
- remove gstlog.h include from gstobject.h, add to .c files
- removed unused refcounting code from gstobject
- small fixes in #includes
- Added Scheduling policy and priority properties to gstthread so that
SCHED_FIFO and SCHED_RR threads can be constructed.
Original commit message from CVS:
remove the "iterate_started" signal and replace it with pre_iterate and post_iterate callbacks.
Apps can now put a lock around an iterate so they can do stuff on the bin from another thread. much nicer than the signal approach.
Original commit message from CVS:
added a "iterate_started" signal which fires before the iterate of a bin. useful for non-threadsafe operations like seek in threaded pipelines
Original commit message from CVS:
- make all _get_list functions return a const reference to original lists and no copy
- redo code accordingly
- fix compile issue in dparam stuff
Original commit message from CVS:
* add notify back to filesrc, it's needed for MVC applications
* remove notify printouts from gst-launch
* cleanup in gst-plugins configure.ac
* some jack updates
* remove SELF_ITERATING flag in favor of SEF_SCHEDULABLE (not a clear name,
but it's what we have for the moment)
* improve parsing of request pad names, no more sscanf
* fixes to the fastscheduler Makefile.am
Original commit message from CVS:
- Remove the propsprivate header file
- Added new API for properties.
- Moved the clock distribution to the scheduler.
- Removed the lock from GstCaps
- Added boxed types for Caps/Props
- Simplified the clock, new systemclock implementation
- Removed deprecated element_info/send_event functions
- First step at exposing more info in the pad_connect functions
- Queue cleanup
- Make the scheduler aware of other schedulers inside it
- Added the _SELF_SCHEDULABLE flag to gstthread
- Removed _get_widget from _utils, changed to new props API
- Make fakesink sync on timestamps when requested
- Removed the offset notify from filesrc
- Added a fast scheduler
- some scheduler cleanups.
Original commit message from CVS:
* s/gst_element_install_std_props/gst_element_class_install_std_props/ -- it just makes more sense that way
* added jack element, doesn't quite work right yet but i didn't want to lose the work -- it does build, register,
and attempt to run though
* imposed some restrictions on the naming of request pads to better allow for reverse parsing
* added '%s' to reverse parsing
* added new bin flag to indicate that it is self-iterating, and some lame code in gst-launch to test it out
* fixen on launch-gui
* added pkg-config stuff for the editor's libs
Original commit message from CVS:
* add two apis:
* gst_element_connect_elements_many (elem1, elem2, ...)
- calls gst_element_connect_elements() in order
- should be called gst_element_connect_many, but we need to rename gst_element_connect_elements first
- simplifies common-case code
* gst_bin_add_many (bin, elem1, ...)
- calls gst_bin_add on all of the elems
- again, simplifying common code
Original commit message from CVS:
Added a first stab at a better clocking system.
It still needs more infrastructure for async notification and custom clock
implementors.
This thing can still deadlock the pipeline.
Original commit message from CVS:
- Added a function to get the currently executing cothread
- Removed some useless includes
- _interrupt now returns a boolean so the behaviour after the interrupt
can be controlled by the scheduler.
- Added a better way to set/get the default scheduler.
- make thread and pipeline get the default scheduler.
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.
Original commit message from CVS:
- Reclaim cothread state even if the cothread was never activated after
the free.
- fix the bin state change when there are no more children.
- use a real method to signal the parent bin of a state change.
- move the state change policy in the scheduler.
- remove the unused and ridiculous state change event
- don't free the cothread state when the element is disabled. only free it
when the element is removed from the scheduler.
Original commit message from CVS:
- added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu>
- some fixes to int2float
- aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at
making automake 1.5 happy (gst now requires automake1.5). It's still not
perfect but it builds.
- Made the schedulers plugable. The default scheduler now lives inside a
plugin.
- Added a new mpeg1/2 parser/demuxer.
- Fixed some compiler warnings in the core libs.
- substantial work to GstThread (hopefully less race conditions). simplified
the code in GstThread a bit. A state change can now also happen in the
thread context.
- reworked the state semantics of a bin. it'll now automatically get the
highest state of its children.
- the autoplugger now nests the threads so that a state change failure of
one thread doesn't make its upstream thread lock.
- GstQueue refuses to go to PLAYING if the sinkpad is not connected. This
way the queue will not wedge in the _get lock.
- GstQueue unlocks its mutexes when going to PAUSED.
- make sure that when all elements in a bin/thread go to PAUSED, the bin
is set to PAUSED too.
- make a parent bin wait for its children to PAUSE before ending the
iteration with FALSE (EOS)
- Some changes to GstPlay to deal with EOS.
- aplied the latest patch from Zeenix to gstrtp.
end result: GstPlay doesn't crash on EOS and the pipeline is now shut down
properly.
Original commit message from CVS:
API docs.
revived _buffer_ref_by_count
fast types for scheduler and bin.
error checking on plugin features;
removed some prototypes that were not implemented (gst_pipeline_iterate
comes to mind)
remove gst_pad_event until we know what it's supposed to do.
remove sinesrc, it wasn't compiles anymore, so...
updates to various elements that used the old event API.
Original commit message from CVS:
This is a megapatch with the following changes:
- moved the gchar *name to GstObject, removed the ones in GstElement and
GstPad.
- moved the parent handling completely into GstObject. This cause *all* of
the plugins to fail (except those that used gst_pad_get_parent)
- rearanged the XML save handling.
- GstObject now has a class function save/restore_thyself.
- GstObject has a generic method gst_object_save_thyself, this makes it
possible to fire a signal wehever a new object is loaded. This is needed
so we can add XML save hooks.
- GstXML API has changed slightly. You now have to create a GstXML object
first before you can actually load something. This makes it possible to
attach a signal to GstXML whenever an object is loaded. I'm not sure we
will keep this interface.
- GstObject can now print the path_string without knowing about the GstPad and
GstElement types.
- Added gst_bin_get_by_name_recurse_up to lookup an element in the current
element hierarchy.
- added XML namespaces to the saved pipelines the namespace is:
http://gstreamer.net/gst-core/1.0/
namespaces are needed to distinguish user generated XML from the core XML.
Note that the plugins still contain a macro GST_OBJECT_PARENT that will be
replaced with gst_pad_get_parent shortly.
Original commit message from CVS:
More EOS changes.
When a bin2 is found inside a bin1, we add the bin2 to the EOS providers
of the bin1. When there is nothing more to schedule in bin1 and bin2 has
fired EOS, bin1 is in EOS.
The queue overrides the EOS notification and calls EOS on the src pad
when the queue is empty and the sink pad is in EOS.
Original commit message from CVS:
A very small change to make eos somewhat work. no inner bins are checked.
When an element fires EOS, the chain with that element is removed from
the scheduler (marked inactive). If all chains are inactive, the bin
fires EOS.
Original commit message from CVS:
Updated the manual and the docs.
Removed the esdsink in gst/elements/ we have a real one not in the
plugins dir.
Added more APIs to query the plugins, types and caps. more fields now
have a getter and a setter. This is needed to make gstreamer wrapper
fiendly.
Added gst_element_disconnect beacuse we also have a gst_element_connect
Original commit message from CVS:
Added code to attempt the scheduling of chained cases. Very simply right
now. Also added code to allow the use_cothread flag of a bin to force all
the chains to use cothreads as well.
Original commit message from CVS:
More massive changes to the scheduling system. Moved the scheduling code
to gstscheduler.[ch], so an child bin can replace the scheduler.
Introduced the concept of chains, which are subsets of the list of managed
elements for a given manager bin, which get scheduled as separate entities.
gst_bin_iterate_func should be pretty much fixed now, the scheduling code
gets to do all the hard work.
Cothreaded case work in the couple tests I've tried, chained is next.
Original commit message from CVS:
Lots of editor changes:
- restucturing of object creation
- loading of xml files
- visual eye candy: color changes on mouseover
- active object is indicated
- property box changes for boolean and enum types
- property box: create properties per element
- show pads/caps in propertybox
added gst_util_get_bool_arg in gstutils.c
added default properties for audiosink
Original commit message from CVS:
Massive scheduling changes (again). Not entirely complete, but getting
closer. Need to think about various scheduling plans that we might want
to produce, and figure out the rules for what is legal, and how to get
the results we need as far as the plan.
Original commit message from CVS:
Header cleanup: try to include as little as possible; this will probably
speed up compilation a bit.
changed the .c files to use #include "..."
Fix for the 'plugins are loaded twice' bug.
Fix 22186: GstObject flags are now used everywhere. Added *_FLAG_LAST so
elements do not use the same flags. Added some padding in the flag enum
for future expansion.
Original commit message from CVS:
Snapshot of work-in-progress do deal with out-of-bin elements. Current
system is very likely going to be ditched due to its complexity and the
fact that it doesn't work right now. More than that, it's generating some
really odd results in my test program, which aren't readily explainable.
If you want to keep a working copy of your working copy (heh), don't update
until this is replaced with a hybrid approach. This approach will be the
beginning of a plan generator that can construct full hybrid schedules
given hints from various places.
Hybrid means that cothreads are used, but there are chunks of the pipeline
that are actually dealt with by chaining. This can improve speed by
reducing cothread switches (which are much cheaper than thread switches,
but still far from free, about 570 cycles on my PIII), but does carry a
complexity burden. Luckily, the structure of GStreamer allows that burden
to live entirely in create_plan. Luck? I think not ;-)
Original commit message from CVS:
Changes to gstreamer-config to include gtk+ libs
manual changes: queues, threads, programs
gsteditor does not crash anymore.
gstpipline new should return a GstElement *
fixed ac3dec for new getbits
fixes to gstreamer-launch
more efficient startup for gstplay.
Original commit message from CVS:
A first (rude) attempt at autoplug.
Autoplugging selects appropriate codecs to connect src to sink, adds
them to the pipeline and connect pads.
Autoplugging will run the typedetect plugin if the src pad has no MIME
type.
No autoplugging is done on the src and sink pads, it's hardcoded:
connect 'src to sink'.
No attempt at creating threads.
No attempt at dynamically autoplugging not yet existing pads.
Changes to (some) plugins to properly set their MIME types.
Original commit message from CVS:
Megapatch, changes which states are available, how they're used, and how
they're set. Also modifies the scheduling system, breaking pulled
buffers. Check mail archives for more details.
