platform/upstream/gst-editing-services.git
9 years agocomposition: Avoid emitting COMMITED like crazy
Mathieu Duponchelle [Fri, 11 Jul 2014 16:29:17 +0000 (18:29 +0200)]
composition: Avoid emitting COMMITED like crazy

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Add an enum to define why we update the stack
Thibault Saunier [Fri, 11 Jul 2014 16:27:25 +0000 (18:27 +0200)]
composition: Add an enum to define why we update the stack

Making the code simpler to follow

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: We are now waiting for caps to restart our task
Mathieu Duponchelle [Fri, 11 Jul 2014 15:48:05 +0000 (17:48 +0200)]
composition: We are now waiting for caps to restart our task

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Remove useless functions to add GSources and add debugging
Thibault Saunier [Fri, 11 Jul 2014 15:31:34 +0000 (17:31 +0200)]
composition: Remove useless functions to add GSources and add debugging

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Remove DONE fixme
Mathieu Duponchelle [Fri, 11 Jul 2014 15:25:44 +0000 (17:25 +0200)]
composition: Remove DONE fixme

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agoghostpad: Do not try to be smarter than possible with seqnum
Thibault Saunier [Fri, 11 Jul 2014 15:23:29 +0000 (17:23 +0200)]
ghostpad: Do not try to be smarter than possible with seqnum

We can have several CAPS event comming at any time and thuse we will
need to rely on elements to handle their seqnum properly as we can not
do a safe guard at our level

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Keep our GSources in a list making their thread safe
Mathieu Duponchelle [Fri, 11 Jul 2014 15:22:24 +0000 (17:22 +0200)]
composition: Keep our GSources in a list making their thread safe

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Concider the last action as DONE when we get a CAPS or SEGMENT
Thibault Saunier [Fri, 11 Jul 2014 14:08:20 +0000 (16:08 +0200)]
composition: Concider the last action as DONE when we get a CAPS or SEGMENT

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: only forward our flush start / stops
Mathieu Duponchelle [Fri, 11 Jul 2014 13:41:50 +0000 (15:41 +0200)]
composition: only forward our flush start / stops

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agoghostpad: remove useless debug
Thibault Saunier [Fri, 11 Jul 2014 12:18:58 +0000 (14:18 +0200)]
ghostpad: remove useless debug

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: cleanup and enhance debug
Mathieu Duponchelle [Fri, 11 Jul 2014 12:17:36 +0000 (14:17 +0200)]
composition: cleanup and enhance debug

Bye Bye STRAM START you were a brave little debug we will miss you.

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: do not flush downstream when updating pipeline ourselves.
Thibault Saunier [Fri, 11 Jul 2014 12:11:21 +0000 (14:11 +0200)]
composition: do not flush downstream when updating pipeline ourselves.

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Set update to TRUE when updating the stack because of EOS
Mathieu Duponchelle [Fri, 11 Jul 2014 10:20:53 +0000 (12:20 +0200)]
composition: Set update to TRUE when updating the stack because of EOS

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Handle seeking current stack while PAUSED
Thibault Saunier [Thu, 10 Jul 2014 16:01:32 +0000 (18:01 +0200)]
composition: Handle seeking current stack while PAUSED

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Add and enhance some debug
Mathieu Duponchelle [Thu, 10 Jul 2014 14:26:48 +0000 (16:26 +0200)]
composition: Add and enhance some debug

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Dot the newly created stacks
Thibault Saunier [Thu, 10 Jul 2014 14:21:31 +0000 (16:21 +0200)]
composition: Dot the newly created stacks

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agognlghostpad: Add some more debugging and fix mistakes in seqnum handling
Mathieu Duponchelle [Thu, 10 Jul 2014 14:17:20 +0000 (16:17 +0200)]
gnlghostpad: Add some more debugging and fix mistakes in seqnum handling

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Factor out a function to seek the current stack
Thibault Saunier [Thu, 10 Jul 2014 13:48:50 +0000 (15:48 +0200)]
composition: Factor out a function to seek the current stack

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Set the seqnum only when receiving the actual seek event
Mathieu Duponchelle [Thu, 10 Jul 2014 13:46:19 +0000 (15:46 +0200)]
composition: Set the seqnum only when receiving the actual seek event

Setting it before calling seek_handling is racy!

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Set the real_eos_seqnum in a dedicated method
Thibault Saunier [Thu, 10 Jul 2014 13:45:19 +0000 (15:45 +0200)]
composition: Set the real_eos_seqnum in a dedicated method

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Reset the real_eos_seqnum when reseting the composition
Mathieu Duponchelle [Thu, 10 Jul 2014 13:43:26 +0000 (15:43 +0200)]
composition: Reset the real_eos_seqnum when reseting the composition

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Release OBJECTS_LOCK when emiting the "commited"
Thibault Saunier [Thu, 10 Jul 2014 13:42:48 +0000 (15:42 +0200)]
composition: Release OBJECTS_LOCK when emiting the "commited"

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Use the private struct directly
Mathieu Duponchelle [Thu, 10 Jul 2014 13:33:23 +0000 (15:33 +0200)]
composition: Use the private struct directly

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agognloperation: Totally clear operations when removing emptying current_bin
Thibault Saunier [Wed, 9 Jul 2014 10:51:36 +0000 (12:51 +0200)]
gnloperation: Totally clear operations when removing emptying current_bin

Removing linked childre leads to weird behaviour, we want to make sure
that all elements are totally clean when they are out the current bin.

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agognlghostpad: Do not forget to set output segment seqnum
Mathieu Duponchelle [Tue, 8 Jul 2014 21:25:09 +0000 (23:25 +0200)]
gnlghostpad: Do not forget to set output segment seqnum

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Fix the build
Thibault Saunier [Tue, 8 Jul 2014 21:00:29 +0000 (23:00 +0200)]
composition: Fix the build

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agotests:composition: Fix some refcounts
Mathieu Duponchelle [Tue, 8 Jul 2014 20:59:11 +0000 (22:59 +0200)]
tests:composition: Fix some refcounts

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Relink nodes *after* setting the seqnum
Thibault Saunier [Tue, 8 Jul 2014 20:58:02 +0000 (22:58 +0200)]
composition: Relink nodes *after* setting the seqnum

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agognlghostpad: Do not tolerate getting seeked when no target is set
Mathieu Duponchelle [Tue, 8 Jul 2014 20:57:22 +0000 (22:57 +0200)]
gnlghostpad: Do not tolerate getting seeked when no target is set

This should not happen anymore

+ Do not set twice events seqnums

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agognlghostpad: Add missing seqnum tweaking
Thibault Saunier [Tue, 8 Jul 2014 20:55:15 +0000 (22:55 +0200)]
gnlghostpad: Add missing seqnum tweaking

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agotests: Use compositor instead of videomixer
Mathieu Duponchelle [Tue, 8 Jul 2014 20:54:31 +0000 (22:54 +0200)]
tests: Use compositor instead of videomixer

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Handle the case were we get an EOS right after a segment
Thibault Saunier [Tue, 8 Jul 2014 20:53:57 +0000 (22:53 +0200)]
composition: Handle the case were we get an EOS right after a segment

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Make basic seeking working
Mathieu Duponchelle [Tue, 8 Jul 2014 11:28:57 +0000 (13:28 +0200)]
composition: Make basic seeking working

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agotests: Start fixing operation tests
Thibault Saunier [Mon, 7 Jul 2014 21:28:43 +0000 (23:28 +0200)]
tests: Start fixing operation tests

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agotests: Fix composition tests
Mathieu Duponchelle [Mon, 7 Jul 2014 21:24:46 +0000 (23:24 +0200)]
tests: Fix composition tests

We can not expect a seek event anymore as we are seeking in READY the elements
themselves

+remove actual sinks

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Implement the logic to PAUSE the task while executing actions
Thibault Saunier [Mon, 7 Jul 2014 21:07:15 +0000 (23:07 +0200)]
composition: Implement the logic to PAUSE the task while executing actions

We need to wait for the pipeline update to be actually finished before we can start another
action. That means that we pause the task until one buffer from the new stack is
outputed.

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Do not forget to set expandables state to NULL when disposing
Mathieu Duponchelle [Mon, 7 Jul 2014 21:08:56 +0000 (23:08 +0200)]
composition: Do not forget to set expandables state to NULL when disposing

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Factor out a function to commit only the values
Thibault Saunier [Mon, 7 Jul 2014 21:01:24 +0000 (23:01 +0200)]
composition: Factor out a function to commit only the values

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Get the current position before actually commiting values on commit
Mathieu Duponchelle [Mon, 7 Jul 2014 20:58:27 +0000 (22:58 +0200)]
composition: Get the current position before actually commiting values on commit

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Add a method that set the current bin to READY
Thibault Saunier [Mon, 7 Jul 2014 20:54:25 +0000 (22:54 +0200)]
composition: Add a method that set the current bin to READY

We need to get the stream lock in some conditions, and thuse send
flush event in those cases.

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Remove useless code
Mathieu Duponchelle [Mon, 7 Jul 2014 20:50:34 +0000 (22:50 +0200)]
composition: Remove useless code

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Clear the old stack when removing children that where used
Thibault Saunier [Mon, 7 Jul 2014 20:33:09 +0000 (22:33 +0200)]
composition: Clear the old stack when removing children that where used

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Remove useless toplevel_seek argiment from activate_new_stack
Mathieu Duponchelle [Mon, 7 Jul 2014 20:27:21 +0000 (22:27 +0200)]
composition: Remove useless toplevel_seek argiment from activate_new_stack

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Rework the seqnum logic to avoid races when setting the new stack seqnum
Thibault Saunier [Mon, 7 Jul 2014 20:25:51 +0000 (22:25 +0200)]
composition: Rework the seqnum logic to avoid races when setting the new stack seqnum

When we were seeking the same stack without a logic that gurantees that we actually
saw the seek with the new seqnum set, we could have ended up with an EOS set with
the right seqnum even if it was actually not the case.

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Add/Remove children in the same order as they were called
Mathieu Duponchelle [Mon, 7 Jul 2014 19:31:01 +0000 (21:31 +0200)]
composition: Add/Remove children in the same order as they were called

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Remove useless flush_start argument from the stop_task method
Thibault Saunier [Mon, 7 Jul 2014 19:28:28 +0000 (21:28 +0200)]
composition: Remove useless flush_start argument from the stop_task method

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Force setting children state to NULL rebfore unrefing them
Mathieu Duponchelle [Sun, 6 Jul 2014 13:46:22 +0000 (15:46 +0200)]
composition: Force setting children state to NULL rebfore unrefing them

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocompositition: Check last stack in the children thread
Thibault Saunier [Fri, 4 Jul 2014 09:05:41 +0000 (11:05 +0200)]
compositition: Check last stack in the children thread

Avoiding to take the OBJECT_LOCK when recieving EOS. The computation is
based on the GstEvent.seqnum to make sure that the EOS we receive
corresponds to the right sequence.

In that patch we tweak seqnums so that they are correctly computed
avoiding to depend on all elements to do it properly as it might pretty
much not be the case!

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Actiually set current_bin state even when not debugging
Mathieu Duponchelle [Fri, 4 Jul 2014 09:11:53 +0000 (11:11 +0200)]
composition: Actiually set current_bin state even when not debugging

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Set *all* children state when going back to READY and then NULL
Thibault Saunier [Thu, 3 Jul 2014 15:42:06 +0000 (17:42 +0200)]
composition: Set *all* children state when going back to READY and then NULL

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Fix toplevel seek event refcounting
Mathieu Duponchelle [Thu, 3 Jul 2014 15:36:01 +0000 (17:36 +0200)]
composition: Fix toplevel seek event refcounting

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Teach the composition to seek same stack
Thibault Saunier [Thu, 3 Jul 2014 14:46:21 +0000 (16:46 +0200)]
composition: Teach the composition to seek same stack

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: The ref the user gave us is our, and we give another to the bin when...
Thibault Saunier [Thu, 3 Jul 2014 14:44:05 +0000 (16:44 +0200)]
composition: The ref the user gave us is our, and we give another to the bin when needed

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Rename commit_pipeline_func to commit_func
Mathieu Duponchelle [Thu, 3 Jul 2014 14:41:42 +0000 (16:41 +0200)]
composition: Rename commit_pipeline_func to commit_func

We do not commit any pipeline, we commit the new state of the composition internals

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Set the state of the internal bin before removing its children
Thibault Saunier [Thu, 3 Jul 2014 12:48:25 +0000 (14:48 +0200)]
composition: Set the state of the internal bin before removing its children

Otherwize when we set the composition state to READY or NULL we can end
up with children in PAUSED state outside of everything

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agotests: Minor cleanup
Mathieu Duponchelle [Thu, 3 Jul 2014 12:34:11 +0000 (14:34 +0200)]
tests: Minor cleanup

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agognlobject: Allow commiting of object that are not inside a composition
Thibault Saunier [Thu, 3 Jul 2014 12:32:44 +0000 (14:32 +0200)]
gnlobject: Allow commiting of object that are not inside a composition

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Empty current_bin on gnl_composition_reset
Mathieu Duponchelle [Thu, 3 Jul 2014 12:32:18 +0000 (14:32 +0200)]
composition: Empty current_bin on gnl_composition_reset

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Remove now useless external_gst_bin_add_remove field
Thibault Saunier [Thu, 3 Jul 2014 12:31:35 +0000 (14:31 +0200)]
composition: Remove now useless external_gst_bin_add_remove field

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Remove now useless pad probes
Mathieu Duponchelle [Thu, 3 Jul 2014 09:36:20 +0000 (11:36 +0200)]
composition: Remove now useless pad probes

Co-Authored by: Thibault Saunier <tsaunier@gnome.org>

9 years agocomposition: Add an internal bin where that contain used children
Thibault Saunier [Wed, 2 Jul 2014 19:01:31 +0000 (21:01 +0200)]
composition: Add an internal bin where that contain used children

Co-Authored by: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>

9 years agocomposition: Add objects to the pending IO list in a GSource
Thibault Saunier [Wed, 2 Jul 2014 15:33:35 +0000 (17:33 +0200)]
composition: Add objects to the pending IO list in a GSource

This way we make sure we do not manipulate our children from another
thread than the dedicated one.

9 years agocomposition: Start implementing seeking in a GSource
Thibault Saunier [Tue, 1 Jul 2014 16:08:32 +0000 (18:08 +0200)]
composition: Start implementing seeking in a GSource

9 years agocomposition: switch to using an action signal to add and remove objects.
Thibault Saunier [Mon, 30 Jun 2014 14:21:30 +0000 (16:21 +0200)]
composition: switch to using an action signal to add and remove objects.

9 years agocomposition: chain up finalize before clering mcontext_lock.
Mathieu Duponchelle [Mon, 30 Jun 2014 14:29:50 +0000 (16:29 +0200)]
composition: chain up finalize before clering mcontext_lock.

9 years agocomposition: Initialize the first stack async
Thibault Saunier [Mon, 30 Jun 2014 13:12:38 +0000 (15:12 +0200)]
composition: Initialize the first stack async

9 years agoFinish fixing tests
Thibault Saunier [Sun, 29 Jun 2014 20:35:34 +0000 (22:35 +0200)]
Finish fixing tests

9 years agotests/common: disconnect commited handler
Mathieu Duponchelle [Sat, 28 Jun 2014 12:44:24 +0000 (14:44 +0200)]
tests/common: disconnect commited handler

9 years agocomposition: Actually commit in on our own thread
Mathieu Duponchelle [Fri, 27 Jun 2014 14:12:12 +0000 (16:12 +0200)]
composition: Actually commit in on our own thread

Avoiding races

9 years agocomposition: start and stop task in init and finalize
Mathieu Duponchelle [Fri, 27 Jun 2014 15:03:44 +0000 (17:03 +0200)]
composition: start and stop task in init and finalize

9 years agotests: unref message correctly
Mathieu Duponchelle [Fri, 27 Jun 2014 15:01:34 +0000 (17:01 +0200)]
tests: unref message correctly

9 years agocomposition: home grown task
Mathieu Duponchelle [Fri, 27 Jun 2014 14:40:19 +0000 (16:40 +0200)]
composition: home grown task

9 years agoDon't be sync silly test
Mathieu Duponchelle [Fri, 27 Jun 2014 14:12:28 +0000 (16:12 +0200)]
Don't be sync silly test

9 years agocomposition: simplify update_pipeline_func
Mathieu Duponchelle [Fri, 27 Jun 2014 13:00:48 +0000 (15:00 +0200)]
composition: simplify update_pipeline_func

9 years agocomposition: Reset the srcpad target when removing the toplevelentry
Thibault Saunier [Fri, 27 Jun 2014 14:26:09 +0000 (16:26 +0200)]
composition: Reset the srcpad target when removing the toplevelentry

9 years agocomposition: Use a GstPad task to run the update pipeline thread
Thibault Saunier [Fri, 27 Jun 2014 10:15:10 +0000 (12:15 +0200)]
composition: Use a GstPad task to run the update pipeline thread

9 years agognlcomposition: Factor out code to deactivate old stack and activate new one
Thibault Saunier [Thu, 26 Jun 2014 21:48:09 +0000 (23:48 +0200)]
gnlcomposition: Factor out code to deactivate old stack and activate new one

9 years agocomposition: Avoid looping using gotos
Thibault Saunier [Thu, 26 Jun 2014 17:00:03 +0000 (19:00 +0200)]
composition: Avoid looping using gotos

9 years agognlcomposition: Use the new _object_block_and_drop_data where appropriate
Thibault Saunier [Thu, 26 Jun 2014 16:41:48 +0000 (18:41 +0200)]
gnlcomposition: Use the new _object_block_and_drop_data where appropriate

Renaming block_object_src_pad to _object_block_and_drop_data

9 years agognlcomposition: factor out some functions
Thibault Saunier [Wed, 25 Jun 2014 17:39:29 +0000 (19:39 +0200)]
gnlcomposition: factor out some functions

9 years agognlcomposition: Factor out the condition of _parent_or_priority_changed
Thibault Saunier [Wed, 25 Jun 2014 17:18:29 +0000 (19:18 +0200)]
gnlcomposition: Factor out the condition of _parent_or_priority_changed

9 years agognlcomposition: Add a function to block object source pad
Thibault Saunier [Wed, 25 Jun 2014 17:17:55 +0000 (19:17 +0200)]
gnlcomposition: Add a function to block object source pad

9 years agognl: Add the srcpad directly to GnlObject
Mathieu Duponchelle [Tue, 24 Jun 2014 11:44:13 +0000 (13:44 +0200)]
gnl: Add the srcpad directly to GnlObject

Starting from now we will not claim that we support GnlObject that have
several source pads as this is
1- Not true at all;
2- the design of priorities in the GnlComposition tree does not allow that;
3- Not very useful in most of the cases and it complexifies quite a lot the code
   in the composition.

Conflicts:
configure.ac
tests/check/Makefile.am

9 years agoRevert "gnlobject: Commit object in READY_TO_PAUSED"
Thibault Saunier [Wed, 25 Jun 2014 13:35:08 +0000 (15:35 +0200)]
Revert "gnlobject: Commit object in READY_TO_PAUSED"

This causes races when seeking, reverting for now even if we will
probably want to bring something like that back.

This reverts commit 3549e745a8f0de3977b83c60e9b447afaf55d8a0.

9 years agognlsource: remove useless "controls_one" field.
Mathieu Duponchelle [Tue, 24 Jun 2014 10:52:24 +0000 (12:52 +0200)]
gnlsource: remove useless "controls_one" field.

9 years agoImport GNL from 978332e7c4c3bba1949421d28b492540ab471450 'Release 1.4.0'
Thibault Saunier [Tue, 21 Oct 2014 08:35:48 +0000 (10:35 +0200)]
Import GNL from 978332e7c4c3bba1949421d28b492540ab471450 'Release 1.4.0'

9 years agotests: timelineedition: Init GES once at the beginning.
Thibault Saunier [Fri, 15 Aug 2014 16:02:36 +0000 (18:02 +0200)]
tests: timelineedition: Init GES once at the beginning.

Avoiding to forget to init in a particular test and failling stupidly

9 years agotest: backgroundsource: Disable tests that rely on nlecomposition internals
Thibault Saunier [Fri, 15 Aug 2014 16:00:24 +0000 (18:00 +0200)]
test: backgroundsource: Disable tests that rely on nlecomposition internals

We can't rely on that, in particular now that it does not actually
add its children all the time but only when it is needed (and that
it has an internal bin where actual things happen).

9 years agotimeline: Add a method to retrieve a pad from a track
Thibault Saunier [Wed, 13 Aug 2014 11:15:02 +0000 (13:15 +0200)]
timeline: Add a method to retrieve a pad from a track

It allows user to easily get the proxied pad from a track.

API:
  + ges_timeline_get_pad_for_track

9 years agotools:launcher: Avoid commiting when we are not done loading the timeline
Thibault Saunier [Mon, 28 Jul 2014 13:26:18 +0000 (15:26 +0200)]
tools:launcher: Avoid commiting when we are not done loading the timeline

9 years agoges: Use the new GNL element message feature to notify async operations
Thibault Saunier [Fri, 25 Jul 2014 22:31:32 +0000 (00:31 +0200)]
ges: Use the new GNL element message feature to notify async operations

9 years agoxml-formatter: message-forward is not something that should be parsed.
Mathieu Duponchelle [Thu, 24 Jul 2014 15:55:35 +0000 (17:55 +0200)]
xml-formatter: message-forward is not something that should be parsed.

9 years agotimeline: handle async start
Mathieu Duponchelle [Fri, 25 Jul 2014 12:47:07 +0000 (14:47 +0200)]
timeline: handle async start

9 years agotests: implement our validate TestManager.
Mathieu Duponchelle [Thu, 23 Oct 2014 19:46:04 +0000 (21:46 +0200)]
tests: implement our validate TestManager.

And make sure it installs alongside the other validate apps.

https://bugzilla.gnome.org/show_bug.cgi?id=739093

9 years agosmart-video-mixer: unref pads_info hash table in dispose
Tim-Philipp Müller [Thu, 30 Oct 2014 15:54:04 +0000 (15:54 +0000)]
smart-video-mixer: unref pads_info hash table in dispose

Before gst_bin_dispose() runs and destroys elements.

9 years agosmart-adder: fix crash in unit test
Tim-Philipp Müller [Thu, 30 Oct 2014 15:20:18 +0000 (15:20 +0000)]
smart-adder: fix crash in unit test

Unref pads_info hash table in dispose instead of
finalize, i.e. before gst_bin_dispose runs and
destroys pads_info->bin (to which the pads_info
does not hold a ref).

9 years agoAutomatic update of common submodule
Sebastian Dröge [Mon, 27 Oct 2014 17:01:56 +0000 (18:01 +0100)]
Automatic update of common submodule

From 84d06cd to 7bb2bce

9 years agotrack: fix indentation
Tim-Philipp Müller [Sun, 26 Oct 2014 20:36:22 +0000 (20:36 +0000)]
track: fix indentation

9 years agotimeline-element: don't leak name string
Tim-Philipp Müller [Sun, 26 Oct 2014 20:35:30 +0000 (20:35 +0000)]
timeline-element: don't leak name string

9 years agotimeline: fix mutex and all_elements hash table leaks
Tim-Philipp Müller [Sun, 26 Oct 2014 20:34:29 +0000 (20:34 +0000)]
timeline: fix mutex and all_elements hash table leaks