platform/upstream/gstreamer.git
10 years agoparse: Use GSlice for allocating and freeing links and chains
Sebastian Dröge [Tue, 14 Jan 2014 12:44:11 +0000 (13:44 +0100)]
parse: Use GSlice for allocating and freeing links and chains

10 years agoparse: Add comment about why we disable the "tracing"
Sebastian Dröge [Tue, 14 Jan 2014 12:42:20 +0000 (13:42 +0100)]
parse: Add comment about why we disable the "tracing"

It did not print anything useful before anyway, everything
was commented out.

Also remove some unneeded struct members.

10 years agoparse-launch: Add some more failing pipelines
Sebastian Dröge [Tue, 14 Jan 2014 12:36:24 +0000 (13:36 +0100)]
parse-launch: Add some more failing pipelines

Also convert some comments about valgrind warnings to
FIXME comments. These were leaking since some time already.

10 years agoparse: Refactor grammar, make it more consistent and fix conflicts
Fabian Kirsch [Fri, 10 Jan 2014 20:10:17 +0000 (21:10 +0100)]
parse: Refactor grammar, make it more consistent and fix conflicts

https://bugzilla.gnome.org/show_bug.cgi?id=710034

10 years agodocs: ensure GstBufferPools shows up as with GObject features
Stefan Sauer [Mon, 13 Jan 2014 19:48:32 +0000 (20:48 +0100)]
docs: ensure GstBufferPools shows up as with GObject features

GstBufferPool is a GstObject, add the _get_type function to the types file.

10 years agotests: baseparse: add test for reverse playback on passthrough
Thiago Santos [Mon, 13 Jan 2014 11:37:37 +0000 (08:37 -0300)]
tests: baseparse: add test for reverse playback on passthrough

Baseparse stores buffers for reverse playback to push on the next
DISCONT, the issue was that it wouldn't ever check for a discont
on passthrough mode as it skips all real parsing. This test
was create to verify this issue and prevent it from happening again

https://bugzilla.gnome.org/show_bug.cgi?id=721941

10 years agotests: baseparse: add basic test for baseparse
Thiago Santos [Sat, 11 Jan 2014 13:49:17 +0000 (10:49 -0300)]
tests: baseparse: add basic test for baseparse

Just a small test to check that basic playback works

10 years agobaseparse: do not accumulate buffers on passthrough mode
Thiago Santos [Fri, 10 Jan 2014 12:10:06 +0000 (09:10 -0300)]
baseparse: do not accumulate buffers on passthrough mode

If on passthrough during reverse playback, do not accumulate buffers as
baseparse will never check for DISCONT flag to push those buffers.
So just push buffers downstream as if it was forward playback.

https://bugzilla.gnome.org/show_bug.cgi?id=721941

10 years agobasesrc: don't confuse GST_PAD_MODE_NONE and PULL
Stefan Sauer [Mon, 6 Jan 2014 20:47:22 +0000 (21:47 +0100)]
basesrc: don't confuse GST_PAD_MODE_NONE and PULL

Use a switch-case to explicitly handle all pad-modes. This way we don't log an error when the pad is not yet activated.

10 years agogstobject: add FIXME and docs for the disabled notify on parent
Stefan Sauer [Thu, 9 Jan 2014 06:56:55 +0000 (07:56 +0100)]
gstobject: add FIXME and docs for the disabled notify on parent

We haven't found a way to re-enable emitting notify and deep-notify for parent
changes. Add a FIXME-2.0 and a doc blob on the property. See #693281.

10 years agobaseparse: do not ignore TIME segments
Thiago Santos [Wed, 8 Jan 2014 19:28:54 +0000 (16:28 -0300)]
baseparse: do not ignore TIME segments

TIME segments are being ignored and a standard initialized
segment is used instead. This causes issues as not properly detecting
reverse playback or not cliping output based on the segment.

This seems to be a regression from one of the GstSegment/GstEvent
redesigns on the 0.10 -> 1.0 transition

10 years agotests: improve check, also check stream-time
Wim Taymans [Wed, 8 Jan 2014 14:47:10 +0000 (15:47 +0100)]
tests: improve check, also check stream-time

10 years agotests: add unit test for segment _offset_running_time()
Wim Taymans [Wed, 8 Jan 2014 14:31:28 +0000 (15:31 +0100)]
tests: add unit test for segment _offset_running_time()

Add a unit test to check that positive and negative offsets are applied
correctly in various cases.

10 years agosegment: take offset into account in _to_position()
Wim Taymans [Wed, 8 Jan 2014 14:23:00 +0000 (15:23 +0100)]
segment: take offset into account in _to_position()

Take the offset into account when converting between running-time and
segment positions.

10 years agopad: use new segment offset method to apply the offset
Wim Taymans [Wed, 8 Jan 2014 13:57:04 +0000 (14:57 +0100)]
pad: use new segment offset method to apply the offset

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=721422

10 years agosegment: add method to offset the segment running-time
Wim Taymans [Wed, 8 Jan 2014 13:54:47 +0000 (14:54 +0100)]
segment: add method to offset the segment running-time

Add a method that can apply an offset to the calculated running-time of
a segment.

10 years agocollectpads: take offset into account for expected segment position
Wim Taymans [Wed, 8 Jan 2014 13:52:04 +0000 (14:52 +0100)]
collectpads: take offset into account for expected segment position

The firt valid segment position is start + offset.
Also add some more debug and a FIXME

10 years agoinfo: debug segment offset field as well
Wim Taymans [Wed, 8 Jan 2014 13:50:29 +0000 (14:50 +0100)]
info: debug segment offset field as well

10 years agomultiqueue: Allow growing a queue if all other queues are not linked
Sebastian Dröge [Wed, 8 Jan 2014 08:53:09 +0000 (09:53 +0100)]
multiqueue: Allow growing a queue if all other queues are not linked

See https://bugzilla.gnome.org/show_bug.cgi?id=719893

10 years agofilesrc: don't try to seek to -1 offset
Wim Taymans [Tue, 7 Jan 2014 15:18:37 +0000 (16:18 +0100)]
filesrc: don't try to seek to -1 offset

The offset can be -1 when we are configured in TIME format. Instead of
failing the seek and erroring, do what and offset of -1 is supposed to
do and simply read from the current offset.

10 years agobasesrc: demote error to warning
Stefan Sauer [Mon, 6 Jan 2014 20:04:32 +0000 (21:04 +0100)]
basesrc: demote error to warning

This is not an error. A subclass returning FALSE for is_seekable() is one way of
saying that we can't operate in pull mode.

10 years agomultiqueue: Fix hanging if shut down while handling a serialized query
YanpingZhang [Fri, 3 Jan 2014 03:47:23 +0000 (11:47 +0800)]
multiqueue: Fix hanging if shut down while handling a serialized query

https://bugzilla.gnome.org/show_bug.cgi?id=721253

10 years agobaseparse: remove pending_segment as it was being misused
Thiago Santos [Thu, 2 Jan 2014 19:22:37 +0000 (16:22 -0300)]
baseparse: remove pending_segment as it was being misused

It wasn't required, instead baseparse was using it to check the media
caps to identify if it was handling audio or video.

The pending_segment was removed and a checked_media boolean
replaced it for a more accurate naming.

https://bugzilla.gnome.org/show_bug.cgi?id=721350

10 years agobaseparse: push pending events before GAP event
Thiago Santos [Thu, 2 Jan 2014 16:43:54 +0000 (13:43 -0300)]
baseparse: push pending events before GAP event

A GAP event is handled as an empty buffer by sinks and they expect
to receive start up events before GAP events (like a segment).

This is important specially if there is a GAP at the beginning of
a stream (before any buffers) so that the segment event can be
pushed downstream before the GAP

https://bugzilla.gnome.org/show_bug.cgi?id=721350

10 years agobaseparse: refactor pending events pushing
Thiago Santos [Thu, 2 Jan 2014 16:41:25 +0000 (13:41 -0300)]
baseparse: refactor pending events pushing

Refactor code repeated 3 times to a common function

https://bugzilla.gnome.org/show_bug.cgi?id=721350

10 years agodesign/part-toc.txt: update design docs
Stefan Sauer [Thu, 2 Jan 2014 19:17:58 +0000 (20:17 +0100)]
design/part-toc.txt: update design docs

10 years agopad: Add unit test for adding/removing blocking probes while a pad is blocked
Sebastian Dröge [Thu, 2 Jan 2014 12:34:52 +0000 (13:34 +0100)]
pad: Add unit test for adding/removing blocking probes while a pad is blocked

And make sure that these new probes are actually called if they should
instead of silently blocking the pad forever.

https://bugzilla.gnome.org/show_bug.cgi?id=721289

10 years agopad: Check if new probes need to be called when adding/removing some
Sebastian Dröge [Thu, 2 Jan 2014 12:33:20 +0000 (13:33 +0100)]
pad: Check if new probes need to be called when adding/removing some

This allows blocking a pad, add a new blocking probe, removing
the first probe and then having the second probe called. Which
could then decide that data-flow should actually continue
instead of blocking now.

https://bugzilla.gnome.org/show_bug.cgi?id=721289

10 years agotee: Remove dyn lock
Sebastian Dröge [Thu, 2 Jan 2014 10:13:27 +0000 (11:13 +0100)]
tee: Remove dyn lock

It was used for pad-alloc in 0.10 but currently is completely unused
and not necessary. All pad access is protected by the tee object lock
and keeping another reference to the current pad.

10 years agotee: Keep another ref to our one and only srcpad around while pushing
Sebastian Dröge [Thu, 2 Jan 2014 10:09:59 +0000 (11:09 +0100)]
tee: Keep another ref to our one and only srcpad around while pushing

A pad probe on that pad might otherwise just release the pad, drop
the last reference and cause great misery.

https://bugzilla.gnome.org/show_bug.cgi?id=721300

10 years agopad: Keep an extra ref of the pad when calling an IDLE probe immediately
Sebastian Dröge [Mon, 30 Dec 2013 18:03:22 +0000 (19:03 +0100)]
pad: Keep an extra ref of the pad when calling an IDLE probe immediately

The callback might destroy the pad.

10 years agopad: Only call IDLE probes if we are actually idle
Sebastian Dröge [Mon, 30 Dec 2013 17:44:24 +0000 (18:44 +0100)]
pad: Only call IDLE probes if we are actually idle

Also only check the data types for non-IDLE probes. When we
are idle, we have no data type obviously.

Previously we were calling IDLE probes during data flow whenever
a non-blocking probe would be called. The pad was usually not idle
at that time.

10 years agonettimeprovider: remove unnecessary NULL check
Tim-Philipp Müller [Mon, 30 Dec 2013 16:10:08 +0000 (16:10 +0000)]
nettimeprovider: remove unnecessary NULL check

Error is never NULL when we break out of the loop.

COVERITY CID 1037151

10 years agoRevert "nettimeprovider: Remove dead code"
Tim-Philipp Müller [Mon, 30 Dec 2013 16:05:47 +0000 (16:05 +0000)]
Revert "nettimeprovider: Remove dead code"

This reverts commit 9649cd4ca19a75d0319117a77a7609e4b8c96533.

This is not right, and it's also not what coverity
is complaining about.

10 years agoqueue: Remove unneeded checks
Edward Hervey [Mon, 30 Dec 2013 15:47:32 +0000 (16:47 +0100)]
queue: Remove unneeded checks

item is guaranteed to be non-null.

COVERITY CID 1037152
COVERITY CID 1037153

10 years agonettimeprovider: Remove dead code
Edward Hervey [Mon, 30 Dec 2013 15:34:08 +0000 (16:34 +0100)]
nettimeprovider: Remove dead code

err is always NULL by the point we reach this line

COVERITY CID 1037151

10 years agotaglist: fix since marker again (1.3 -> 1.4)
Stefan Sauer [Mon, 30 Dec 2013 13:47:19 +0000 (14:47 +0100)]
taglist: fix since marker again (1.3 -> 1.4)

Since markers should point to the appropriate stable version.

10 years agotaglist: fix since-marker in docs (we're in 1.3.X)
Stefan Sauer [Mon, 30 Dec 2013 13:40:40 +0000 (14:40 +0100)]
taglist: fix since-marker in docs (we're in 1.3.X)

10 years agotaglist: add a tag for midi base note numbers
Stefan Sauer [Mon, 30 Dec 2013 13:22:37 +0000 (14:22 +0100)]
taglist: add a tag for midi base note numbers

Audio files containing sampled instruments can have metadata describing the note
that was played on the instrument.

10 years agopad: Don't ignore probe callback return value when immediately calling IDLE probe
Sebastian Dröge [Mon, 30 Dec 2013 09:01:39 +0000 (10:01 +0100)]
pad: Don't ignore probe callback return value when immediately calling IDLE probe

https://bugzilla.gnome.org/show_bug.cgi?id=721096

10 years agoscripts: git-update.sh: fix for non-master branches
Pedro Côrte-Real [Sun, 29 Dec 2013 14:06:55 +0000 (14:06 +0000)]
scripts: git-update.sh: fix for non-master branches

Pull from tracking branch instead of origin/master, so
that this works with e.g. 1.2 as well.

10 years agoparse: Use GLib malloc/free/realloc functions
Sebastian Dröge [Fri, 27 Dec 2013 11:55:02 +0000 (12:55 +0100)]
parse: Use GLib malloc/free/realloc functions

https://bugzilla.gnome.org/show_bug.cgi?id=720100

10 years agoscripts: create-uninstalled-setup: re-use existing master branch if it exists
Tim-Philipp Müller [Thu, 26 Dec 2013 12:16:26 +0000 (12:16 +0000)]
scripts: create-uninstalled-setup: re-use existing master branch if it exists

When creating separate checkout for non-master branches.

10 years agodocs: fix memory leak of appsink example in manual
Erik Andresen [Thu, 26 Dec 2013 10:39:27 +0000 (11:39 +0100)]
docs: fix memory leak of appsink example in manual

https://bugzilla.gnome.org/show_bug.cgi?id=721076

10 years agoAutomatic update of common submodule
Tim-Philipp Müller [Sun, 22 Dec 2013 22:33:12 +0000 (22:33 +0000)]
Automatic update of common submodule

From dbedaa0 to d48bed3

10 years agopo: update for string changes
Tim-Philipp Müller [Sun, 22 Dec 2013 22:04:37 +0000 (22:04 +0000)]
po: update for string changes

10 years agopo: set gettext domain in Makevars so we don't have to patch the generated Makefile...
Tim-Philipp Müller [Sun, 22 Dec 2013 21:56:03 +0000 (21:56 +0000)]
po: set gettext domain in Makevars so we don't have to patch the generated Makefile.in.in

https://bugzilla.gnome.org/show_bug.cgi?id=705455

10 years agoutils: Add round down 128 macro for completeness
Wim Taymans [Fri, 20 Dec 2013 13:41:06 +0000 (14:41 +0100)]
utils: Add round down 128 macro for completeness

10 years agoutils: Add round up 128 macro
Nicolas Dufresne [Wed, 13 Nov 2013 22:06:23 +0000 (17:06 -0500)]
utils: Add round up 128 macro

10 years agofunnel: Setting the PROXY_CAPS flag on the srcpad does not make much sense
Sebastian Dröge [Fri, 20 Dec 2013 13:30:22 +0000 (14:30 +0100)]
funnel: Setting the PROXY_CAPS flag on the srcpad does not make much sense

funnel outputs whatever one of the upstreams currently outputs, a caps
query to a random upstream does not give the right answer here.

10 years agofunnel: Proxy CAPS and ALLOCATION queries
Sebastian Dröge [Fri, 20 Dec 2013 10:37:53 +0000 (11:37 +0100)]
funnel: Proxy CAPS and ALLOCATION queries

10 years agomultiqueue: post 100% buffering if single queue is not linked
Thiago Santos [Mon, 18 Nov 2013 13:46:00 +0000 (10:46 -0300)]
multiqueue: post 100% buffering if single queue is not linked

This makes buffering stop in case a stream switch happens. This is
important for adaptive streams that can disable not-linked streams
to avoid consuming the network bandwidth.

https://bugzilla.gnome.org/show_bug.cgi?id=719575

10 years agodocs: fix project links
William Jon McCann [Wed, 18 Dec 2013 10:19:46 +0000 (05:19 -0500)]
docs: fix project links

https://bugzilla.gnome.org/show_bug.cgi?id=720665

10 years agobasesrc: use segment start if DTS for first buffer is unset
David Svensson Fors [Tue, 10 Dec 2013 14:53:54 +0000 (15:53 +0100)]
basesrc: use segment start if DTS for first buffer is unset

https://bugzilla.gnome.org/show_bug.cgi?id=720199

10 years agogst-uninstalled: add gstreamer-vaapi paths
Matthieu Bouron [Thu, 12 Dec 2013 17:17:40 +0000 (17:17 +0000)]
gst-uninstalled: add gstreamer-vaapi paths

https://bugzilla.gnome.org/show_bug.cgi?id=720337

10 years agotests: add unit test for bitreader, bytereader and bytewriter with no inlining used
Tim-Philipp Müller [Sat, 14 Dec 2013 21:20:45 +0000 (21:20 +0000)]
tests: add unit test for bitreader, bytereader and bytewriter with no inlining used

10 years agobytereader: add inline variant of gst_byte_reader_init()
Tim-Philipp Müller [Sat, 14 Dec 2013 18:38:41 +0000 (18:38 +0000)]
bytereader: add inline variant of gst_byte_reader_init()

10 years agodocs: fix docs for gst_queue_array_peek_head()
Tim-Philipp Müller [Sat, 14 Dec 2013 18:31:38 +0000 (18:31 +0000)]
docs: fix docs for gst_queue_array_peek_head()

10 years agoparse: Don't define yyscan_t twice
Sebastian Dröge [Sat, 14 Dec 2013 18:08:35 +0000 (19:08 +0100)]
parse: Don't define yyscan_t twice

https://bugzilla.gnome.org/show_bug.cgi?id=720316

10 years agodocs: moving plugins: minor 0.10 -> 1.0 fix
Tim-Philipp Müller [Fri, 13 Dec 2013 22:51:32 +0000 (22:51 +0000)]
docs: moving plugins: minor 0.10 -> 1.0 fix

Spotted by Jay Fenlason

10 years agoqueue: don't ignore event return value
Wim Taymans [Wed, 11 Dec 2013 13:42:34 +0000 (14:42 +0100)]
queue: don't ignore event return value

Pass the event return value upstream.
Remove strange goto construct.

10 years agogst-launch: Handle taglist copy failure
Edward Hervey [Tue, 10 Dec 2013 23:30:03 +0000 (18:30 -0500)]
gst-launch: Handle taglist copy failure

If we couldn't copy the tags, just return instead of trying to use bogus
values.

10 years agogst-inspect: Index features are no more
Edward Hervey [Tue, 10 Dec 2013 23:25:22 +0000 (18:25 -0500)]
gst-inspect: Index features are no more

So remove code that will never be used

10 years agogstvalue: Fix comparision of double range
Edward Hervey [Tue, 10 Dec 2013 22:53:24 +0000 (17:53 -0500)]
gstvalue: Fix comparision of double range

Checking twice the lower bound is great (you never know, it might change
between the two calls by someone using emacs butterfly-mode), but it's a bit
more useful to check the higher bound are also identical.

Detected by Coverity

10 years agocontroller: Fix out-of-bounds detection
Edward Hervey [Tue, 10 Dec 2013 22:09:07 +0000 (17:09 -0500)]
controller: Fix out-of-bounds detection

We want to abort if we higher than the maximum *OR* lower than the minimum
accepted value.

Detected by Coverity.

10 years agotests: add unit test for registering the same category twice
Tim-Philipp Müller [Sat, 7 Dec 2013 19:04:16 +0000 (19:04 +0000)]
tests: add unit test for registering the same category twice

10 years agoinfo: return existing category if a debug category is registered twice
Víctor Manuel Jáquez Leal [Sat, 7 Dec 2013 18:32:58 +0000 (19:32 +0100)]
info: return existing category if a debug category is registered twice

If a category with the same name is found when creating a new
one, the found category is returned instead of an invalid pointer.
Fixes issue with gst-vaapi (which uses an internal copy of the
codec parsers) caused by commit ccba9130.

https://bugzilla.gnome.org/show_bug.cgi?id=720036

10 years agodocs: Fix typos in function/object descriptions
Sebastian Rasmussen [Sat, 7 Dec 2013 14:38:19 +0000 (15:38 +0100)]
docs: Fix typos in function/object descriptions

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=720029

10 years agoFix some typos in code comments and debug messages
Sebastian Rasmussen [Sat, 7 Dec 2013 14:40:32 +0000 (15:40 +0100)]
Fix some typos in code comments and debug messages

https://bugzilla.gnome.org/show_bug.cgi?id=720029

10 years agodocs: clarify encoding of strings in GstStructures and taglists
Stewart Brodie [Fri, 6 Dec 2013 20:50:19 +0000 (20:50 +0000)]
docs: clarify encoding of strings in GstStructures and taglists

https://bugzilla.gnome.org/show_bug.cgi?id=709262

10 years agobasesrc: Set format to TIME if do-timestamp is TRUE
Olivier Crête [Wed, 3 Jul 2013 00:27:59 +0000 (20:27 -0400)]
basesrc: Set format to TIME if do-timestamp is TRUE

https://bugzilla.gnome.org/show_bug.cgi?id=702842

10 years agotools: gst-launch: don't try to remove already-removed GSource from main loop
Tim-Philipp Müller [Thu, 5 Dec 2013 00:26:13 +0000 (00:26 +0000)]
tools: gst-launch: don't try to remove already-removed GSource from main loop

It's considered a programming error in recent GLib versions now.
We may already have removed the source by returning FALSE from
the callback if it was fired. Fixes warning with newer GLibs
when interrupting a pipeline with Control-C.

10 years agoinfo: Make sure the same category is not added twice
Olivier Crête [Wed, 4 Dec 2013 22:35:18 +0000 (17:35 -0500)]
info: Make sure the same category is not added twice

10 years agoinfo: Protect __categories list in get_category with lock too
Olivier Crête [Wed, 4 Dec 2013 22:35:02 +0000 (17:35 -0500)]
info: Protect __categories list in get_category with lock too

10 years agodocs: add missing files for distribution
Sebastian Rasmussen [Tue, 3 Dec 2013 23:10:36 +0000 (00:10 +0100)]
docs: add missing files for distribution

 * add some documentation files in docs/design
 * add docs/list-ulink.xsl so check in docs/manual works

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=719814

10 years agopad: add ACCEPT_INTERCEPT flag
Wim Taymans [Tue, 3 Dec 2013 20:46:19 +0000 (21:46 +0100)]
pad: add ACCEPT_INTERCEPT flag

Make a new flag on the pad that tweaks the default behaviour of the
accept-caps function. By default it will check for a subset of the
query-caps result but this is not always desirable. The query-caps
result contains all the constraints to make a good caps decision
upstream but sometimes, like for parsers, not all the constrained caps
fields are known upstream and then a subset check would fail. Switching
to an intersection makes this work again.

See https://bugzilla.gnome.org/show_bug.cgi?id=705024
    https://bugzilla.gnome.org/show_bug.cgi?id=677401

10 years agomultiqueue: Wake up on reconfigure event
Olivier Crête [Tue, 3 Dec 2013 03:22:36 +0000 (22:22 -0500)]
multiqueue: Wake up on reconfigure event

After patch bda406c4, the state of the singlequeue was set to OK, but nothing
would then wake up the thread, as the other wakeup functions only look at
singlequeues that are marked as having received as not-linked.

https://bugzilla.gnome.org/show_bug.cgi?id=708200

10 years agodocs: add missing docs, fixing doc errors
Sebastian Rasmussen [Sat, 30 Nov 2013 11:15:37 +0000 (12:15 +0100)]
docs: add missing docs, fixing doc errors

 * add many missing declarations to sections
 * GstController has been removed, update docs
 * skip GstIndex when generating documentation
 * rephrase so gtkdoc doesn't imagine return value
 * add missing argument description for gst_context_new()
 * document GstOutputSelectorPadNegotiationMode and move to header-file

https://bugzilla.gnome.org/show_bug.cgi?id=719614

10 years agodocs: cosmetic changes in references/decriptions
Sebastian Rasmussen [Sat, 30 Nov 2013 13:52:40 +0000 (14:52 +0100)]
docs: cosmetic changes in references/decriptions

 * fix typo GstBufferFlag -> GstBufferFlags
 * fix typo GstFeatures -> GstCapsFeatures
 * fix typo GstAllocatorParams -> GstAllocationParams
 * fix typo GstContrlSources -> GstControlSource
 * do not refer to gstcheck as an object
 * make references gtk_init() and tcase_set_timeout() not be references
 * gst_element_get_pad() renamed gst_element_get_static_pad()
 * gst_clock_id_wait_async_full() renamed gst_clock_id_wait_async()
 * _drop_element() is really gst_queue_array_drop_element()
 * gst_pad_accept_caps() was removed, do not refer to it
 * separate GST_META_TAG_MEMORY_STR declaration from description
 * do not describe removed gst_collect_pads_collect()
 * correctly link to GstElementClass' virtual set_context()

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=719614

10 years agoparse: Manually insert priv_gst_parse_yyget/set_column prototypes for older flex
Olivier Crête [Fri, 29 Nov 2013 19:00:35 +0000 (14:00 -0500)]
parse: Manually insert priv_gst_parse_yyget/set_column prototypes for older flex

Older versions of flex (before 2.5.36) don't add the prototype, so it must
be added manually. We can't check by the version number, because Debian/Ubuntu
patched it into their 2.5.35 at some point.

10 years agogstpad: drop assertion on gst_pad_peer_query_position
Thiago Santos [Tue, 19 Nov 2013 14:41:54 +0000 (11:41 -0300)]
gstpad: drop assertion on gst_pad_peer_query_position

It is a 'both' query, so it can be sent both ways

10 years agoinputselector: handle gap events
Thiago Santos [Mon, 18 Nov 2013 21:11:56 +0000 (18:11 -0300)]
inputselector: handle gap events

Use gap events to advance the selector's pad position.

This is relevant to keep sync_streams mode working when one of the
streams doesn't have data all the time.

10 years agoRevert "ghostpad: copy sticky events to SRC ghostpads"
Wim Taymans [Fri, 29 Nov 2013 16:02:41 +0000 (17:02 +0100)]
Revert "ghostpad: copy sticky events to SRC ghostpads"

This reverts commit 8162a583a4dd68582bf186e2e47a8f0d68fa1980.

Automatically copying the sticky events makes it impossible for apps
and elements to filter the events with event probes. This causes
regressions (See #719437). The best option is to let the app/element
copy and filter the events themselves after the ghostpad target is
set.

10 years agoparse: fix segfaulting prototype-mismatch
Fabian Kirsch [Tue, 19 Nov 2013 14:03:35 +0000 (15:03 +0100)]
parse: fix segfaulting prototype-mismatch

Now YYDEBUG is always set, so check it's value

https://bugzilla.gnome.org/show_bug.cgi?id=712679

10 years agonetclock: Fix docstring for round-trip-limit and uninit access warning.
Jan Schmidt [Wed, 27 Nov 2013 07:32:22 +0000 (18:32 +1100)]
netclock: Fix docstring for round-trip-limit and uninit access warning.

Fix a typo in a doc string - the property is round-trip-limit, not
roundtrip-limit.

Remove a bogus GST_WARNING that can print an uninitialised variable
and is redundant anyway.

10 years agonetclock: Add round-trip-limit parameter
Carlos Rafael Giani [Tue, 26 Nov 2013 10:56:46 +0000 (11:56 +0100)]
netclock: Add round-trip-limit parameter

Sometimes, packets might take a very long time to return. Such packets
usually are way too late and destabilize the regression with their
obsolete data. On Wi-Fi, round-trips of over 7 seconds have been observed.

If the limit is set to a nonzero value, packets with a round-trip period
larger than the limit are ignored.

Signed-off-by: Carlos Rafael Giani <dv@pseudoterminal.org>
https://bugzilla.gnome.org/show_bug.cgi?id=712385

10 years agoquery: Fix gi annotations of gst_structure_new_custom()
Olivier Crête [Tue, 26 Nov 2013 00:04:38 +0000 (19:04 -0500)]
query: Fix gi annotations of gst_structure_new_custom()

10 years agonetclock: Fix C99 comment
Jan Schmidt [Mon, 25 Nov 2013 15:43:54 +0000 (02:43 +1100)]
netclock: Fix C99 comment

10 years agonetclock: Implement rolling-average filter on observations.
Jan Schmidt [Mon, 25 Nov 2013 15:17:36 +0000 (02:17 +1100)]
netclock: Implement rolling-average filter on observations.

Keep a rolling average of the round trip time for network clock
observations, favouring shorter round trips as being more accurate.

Don't pass any clock observation to the clock slaving if it has a
round-trip time greater than 2 times the average.

Actual shifts in the network topology will be noticed after some
time, as the rolling average incorporates the new round trip times.

10 years agobasesink: Add debug into gst_base_sink_default_query() for accept_caps
Jan Schmidt [Mon, 25 Nov 2013 09:33:42 +0000 (20:33 +1100)]
basesink: Add debug into gst_base_sink_default_query() for accept_caps

10 years agogst-launch: exit with an error code when an error occured
Philippe Normand [Thu, 14 Nov 2013 14:32:59 +0000 (15:32 +0100)]
gst-launch: exit with an error code when an error occured

If the pipeline failed to pre-roll or the user interrupted the
execution then set the exit code to a positive value.

https://bugzilla.gnome.org/show_bug.cgi?id=712300

10 years agogstutils: Escape stream id format in comments
Sebastian Rasmussen [Fri, 22 Nov 2013 00:35:18 +0000 (01:35 +0100)]
gstutils: Escape stream id format in comments

These must be escaped for gtk-doc to parse the comments without warnings.

https://bugzilla.gnome.org/show_bug.cgi?id=714989

10 years agogst: g_memmove() is deprecated
Tim-Philipp Müller [Thu, 21 Nov 2013 15:04:04 +0000 (15:04 +0000)]
gst: g_memmove() is deprecated

Just use plain memmove(), g_memmove() is deprecated in
recent GLib versions.

https://bugzilla.gnome.org/show_bug.cgi?id=712811

10 years agoghostpad: copy sticky events to SRC ghostpads
Wim Taymans [Thu, 21 Nov 2013 13:13:16 +0000 (14:13 +0100)]
ghostpad: copy sticky events to SRC ghostpads

Update the sticky events on SRC ghostpads when retargeting. This ensures
that the ghostpad has the exect same sticky events as the target pad. We
don't want to do this for SINK ghostpads, they got the events from
downstream and we don't want to overwrite them with the target pad
events.

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=707621

10 years agopad: move debug function closer to the enum it debugs
Wim Taymans [Thu, 21 Nov 2013 11:28:00 +0000 (12:28 +0100)]
pad: move debug function closer to the enum it debugs

10 years agopluginloader: check read/write before closed
Wim Taymans [Mon, 18 Nov 2013 20:39:54 +0000 (21:39 +0100)]
pluginloader: check read/write before closed

first try to read or write on the socket before checking the closed state. This
makes sure we handle all data on the socket before erroring out.

10 years agopoll: improve debug
Wim Taymans [Mon, 18 Nov 2013 20:37:06 +0000 (21:37 +0100)]
poll: improve debug

So that we can see the return values of functions in the log.

10 years agotests: fix GstBus unit test with latest GLib
Tim-Philipp Müller [Mon, 18 Nov 2013 15:28:32 +0000 (15:28 +0000)]
tests: fix GstBus unit test with latest GLib

g_source_remove() works on the default main context, and
we're doing things with a custom context. Fixes warning
with newer GLib versions.

10 years agodocs: cosmetic since marker fixes
Tim-Philipp Müller [Sat, 16 Nov 2013 12:24:56 +0000 (12:24 +0000)]
docs: cosmetic since marker fixes

10 years agobaseparse: ensure to preserve upstream timestamps
Mark Nauwelaerts [Sat, 16 Nov 2013 14:17:57 +0000 (15:17 +0100)]
baseparse: ensure to preserve upstream timestamps

... rather than have subclass coming up with an internally parsed one.

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=707230