platform/upstream/gstreamer.git
9 years agoh264parse: expose compatible profiles to downstream
Sreerenj Balachandran [Tue, 2 Dec 2014 08:10:39 +0000 (10:10 +0200)]
h264parse: expose compatible profiles to downstream

Some video bitstreams report a too restrictive set of profiles. If a video
decoder was to strictly follow the indicated profile, it wouldn't support that
stream, whereas it could in theory and in practice. So we should relax the
profile restriction for allowing the decoder to get connected with parser.

https://bugzilla.gnome.org/show_bug.cgi?id=739992

9 years agodshowvideosink: Port to 1.0 and new GLib threading API
Chun-wei Fan [Wed, 1 May 2013 03:17:02 +0000 (11:17 +0800)]
dshowvideosink: Port to 1.0 and new GLib threading API

This updates the dshowvideosink to work with the GStreamer 1.0.x APIs, and
avoids the use of deprecated GLib threading API that are now used since
GLib 2.32+

https://bugzilla.gnome.org/show_bug.cgi?id=699364

9 years agomssdemux: add live support
Thiago Santos [Thu, 4 Dec 2014 04:46:43 +0000 (01:46 -0300)]
mssdemux: add live support

Implement the functions of adaptivedemux to have live support
enabled. This allows mssdemux to refresh the Manifest periodically

9 years agodshowdecwrapper: Port to 1.x
Matthew Bader [Thu, 4 Dec 2014 20:52:44 +0000 (15:52 -0500)]
dshowdecwrapper: Port to 1.x

https://bugzilla.gnome.org/show_bug.cgi?id=740945

9 years agovcdsrc: port to 1.
Reynaldo H. Verdejo Pinochet [Mon, 1 Dec 2014 19:40:24 +0000 (16:40 -0300)]
vcdsrc: port to 1.

https://bugzilla.gnome.org/show_bug.cgi?id=547661

9 years agomssdemux: fix getting current position in segments when after last one
Thiago Santos [Thu, 4 Dec 2014 15:21:04 +0000 (12:21 -0300)]
mssdemux: fix getting current position in segments when after last one

For live streams this makes a difference when resyncing after a
Manifest refresh, otherwise it doesn't know where to restart

9 years agomssdemux: support manifests with repetition fragments
Thiago Santos [Wed, 3 Dec 2014 02:27:29 +0000 (23:27 -0300)]
mssdemux: support manifests with repetition fragments

Read the "r" attribute from fragments to support fragments nodes
that use repetition to have a shorter Manifest xml.

Instead of doing:
<c d="100" />
<c d="100" />

You can use:
<c d="100" r="2" />

9 years agomssdemux: port to adaptivedemux base class
Thiago Santos [Sat, 22 Nov 2014 04:01:50 +0000 (01:01 -0300)]
mssdemux: port to adaptivedemux base class

9 years agodshowsrcwrapper: Fixed null caps comparison
Matthew Bader [Wed, 3 Dec 2014 21:13:17 +0000 (16:13 -0500)]
dshowsrcwrapper: Fixed null caps comparison

https://bugzilla.gnome.org/show_bug.cgi?id=741086

9 years agohlssink: remove unnecessary title on EXTINF tag
Flávio Ribeiro [Thu, 4 Dec 2014 04:12:08 +0000 (02:12 -0200)]
hlssink: remove unnecessary title on EXTINF tag

According to the HLS spec the remainder of the line following
the comma on EXTINF tag is not required. This patch removes
the fake title and saves some bytes on the playlist.

https://bugzilla.gnome.org/show_bug.cgi?id=741096

9 years agogl: Rename gst-launch-0.10 to gst-launch-1.0 in docs
Ramprakash Jelari [Thu, 4 Dec 2014 15:07:51 +0000 (20:37 +0530)]
gl: Rename gst-launch-0.10 to gst-launch-1.0 in docs

9 years agodecklink: Fix make distcheck
Sebastian Dröge [Thu, 4 Dec 2014 14:42:55 +0000 (15:42 +0100)]
decklink: Fix make distcheck

And properly dist all headers and source files.

9 years agoadaptivedemux: release threads waiting for manifest update on shutdown
Thiago Santos [Wed, 3 Dec 2014 21:31:06 +0000 (18:31 -0300)]
adaptivedemux: release threads waiting for manifest update on shutdown

For live streams, also unblock threads that might be waiting for a manifest
update when stopping the pipeline.

Also add some more debug messages

9 years agoadaptivedemux: use gst_task_stop instead of gst_task_pause
Thiago Santos [Wed, 3 Dec 2014 16:52:07 +0000 (13:52 -0300)]
adaptivedemux: use gst_task_stop instead of gst_task_pause

To avoid race conditions with gst_task_stop(); gst_task_join() with
another thread doing gst_task_pause(), the joining thread would be
waiting for the task to stop but it would never happen. So just
use gst_task_stop() everywhere to prevent more mutexes

9 years agoadaptivedemux: fix deadlock when stopping streams
Thiago Santos [Wed, 3 Dec 2014 15:48:42 +0000 (12:48 -0300)]
adaptivedemux: fix deadlock when stopping streams

Remember to unlock the manifest's lock when leaving the download task.

9 years agodecklink: Fix compiler warnings with gcc in the dispatcher
Sebastian Dröge [Wed, 3 Dec 2014 15:01:11 +0000 (16:01 +0100)]
decklink: Fix compiler warnings with gcc in the dispatcher

9 years agoglsyncmeta: add G_END_DECLS
Philippe Normand [Wed, 3 Dec 2014 14:30:43 +0000 (15:30 +0100)]
glsyncmeta: add G_END_DECLS

9 years agodecklink: Update Windows SDK headers to 10.3.1
Sebastian Dröge [Wed, 3 Dec 2014 14:07:16 +0000 (15:07 +0100)]
decklink: Update Windows SDK headers to 10.3.1

No idea where the DecklinkAPIDispatch.cpp comes from on Windows,
but this should still work. Will just become a problem once we
use other parts of the API.

9 years agodecklink: Update OS X SDK headers to 10.3.1
Sebastian Dröge [Wed, 3 Dec 2014 13:42:43 +0000 (14:42 +0100)]
decklink: Update OS X SDK headers to 10.3.1

9 years agodecklink: Update Linux SDK headers to 10.3.1
Sebastian Dröge [Wed, 3 Dec 2014 13:39:13 +0000 (14:39 +0100)]
decklink: Update Linux SDK headers to 10.3.1

9 years agoadaptivedemux: remove unused error variable
Thiago Santos [Wed, 3 Dec 2014 13:06:09 +0000 (10:06 -0300)]
adaptivedemux: remove unused error variable

CID #1256556

9 years agoadaptivedemux: remove more get_duration checks
Thiago Santos [Wed, 3 Dec 2014 12:45:58 +0000 (09:45 -0300)]
adaptivedemux: remove more get_duration checks

Subclasses must implement get_duration function. If they do not know
the duration they can return GST_CLOCK_TIME_NONE

9 years agoadaptivedemux: all subclasses must implement get_duration
Thiago Santos [Wed, 3 Dec 2014 12:40:34 +0000 (09:40 -0300)]
adaptivedemux: all subclasses must implement get_duration

CID #1256557

9 years agoadaptivedemux: make sure to always stop the stream on EOS
Thiago Santos [Wed, 3 Dec 2014 02:29:20 +0000 (23:29 -0300)]
adaptivedemux: make sure to always stop the stream on EOS

For On-Demand streams, always stop the tasks for streams that reached
EOS

9 years agoglcontext: try cgl before glx
Julien Isorce [Wed, 3 Dec 2014 08:02:58 +0000 (08:02 +0000)]
glcontext: try cgl before glx

It was already done by commit
f506e80686eedc66912583db11914ec3f713b478
but it has been broken by commit
45ec777cea0e69b2597d1ec02b18ec31d618c920

9 years agotests: add glimagesink unit test to .gitignore
Julien Isorce [Wed, 3 Dec 2014 07:46:50 +0000 (07:46 +0000)]
tests: add glimagesink unit test to .gitignore

9 years agodashdemux: fix double unref of object
Thiago Santos [Tue, 2 Dec 2014 18:52:54 +0000 (15:52 -0300)]
dashdemux: fix double unref of object

the datetime is already released by the end of the function

9 years agodashdemux: Find the current period in a live stream
David Waring [Fri, 10 Oct 2014 10:24:08 +0000 (11:24 +0100)]
dashdemux: Find the current period in a live stream

Start from the correct period on a live stream

https://bugzilla.gnome.org/show_bug.cgi?id=737421

9 years agodashdemux: Fix period selection for live streams
David Waring [Tue, 2 Dec 2014 10:06:00 +0000 (10:06 +0000)]
dashdemux: Fix period selection for live streams

Fix period selection and properly error out when update cannot be done

https://bugzilla.gnome.org/show_bug.cgi?id=725298

9 years agompegtbase: Improve last PCR detection
Edward Hervey [Wed, 9 Jul 2014 08:11:40 +0000 (10:11 +0200)]
mpegtbase: Improve last PCR detection

When dealing with random-access content (such as files), we initially
search for the last PCR in order to figure out duration and to handle
other position estimation such as those used in seeking.

Previously, the code looking for that last PCR would search in the last
640kB of the file going forward, and stop at the first PCR encountered.
The problem with that was two-fold:
* It wouldn't really be the last PCR (it would be the first one within
  those last 640kB. In case of VBR files, this would put off duration
  and seek code slightly.
* It would fail on files with bitrates higher than 52Mbit/s (not common)

Instead this patch modifies that code by:
* Scanning over the last 2048kB (allows to cope with streams up to 160Mbit/s)
* Starts by the end of the file, going over chunks of 300 MPEG-TS packets
* Doesn't stop at the first PCR detected in a chunk, but instead records all
  of them, and only stop searching if there was "at least" one PCR within
  that chunk

This should improve duration reporting and seeking operations on VBR files

https://bugzilla.gnome.org/show_bug.cgi?id=708532

9 years agotsdemux: Only activate warning for real streams
Edward Hervey [Mon, 1 Dec 2014 10:47:29 +0000 (11:47 +0100)]
tsdemux: Only activate warning for real streams

streams with stream_type 0xff are PCR-only streams, it's normal not
to have a pad for them.

9 years agox265enc: add qp property
Jürgen Slowack [Mon, 1 Dec 2014 11:14:20 +0000 (12:14 +0100)]
x265enc: add qp property

QP property enables CQP mode and sets the QP for P slices.

9 years agogl/cocoa: Fix example on Mac OS X 10.10
Thijs Vermeir [Mon, 1 Dec 2014 09:59:32 +0000 (10:59 +0100)]
gl/cocoa: Fix example on Mac OS X 10.10

Using NSApp directly seems to confuse something, as the compiler
was expecting an id<NSFileManagerDelegate>. Switched to using
[NSApplication sharedApplication], and specified the delegate
protocol on the window class as well.

Similar to https://bugzilla.gnome.org/show_bug.cgi?id=738740

9 years agogst-libs: Fix dependency build
Edward Hervey [Mon, 1 Dec 2014 09:27:43 +0000 (10:27 +0100)]
gst-libs: Fix dependency build

adaptivedemux depends on uridownloader and video depends on base

9 years agoconfigure: Do comparisons in a portable way with = instead of ==
Thomas Klausner [Mon, 1 Dec 2014 08:48:41 +0000 (09:48 +0100)]
configure: Do comparisons in a portable way with = instead of ==

https://bugzilla.gnome.org/show_bug.cgi?id=740953

9 years agoadaptivedemux: reworking live manifest update wait
Thiago Santos [Mon, 1 Dec 2014 00:54:38 +0000 (21:54 -0300)]
adaptivedemux: reworking live manifest update wait

Check if the stream is live before checking if it is EOS as a live
stream might be considered EOS when it just needs to wait for a manifest
update to proceed with the next fragments

9 years agoadaptivedemux: fix manifest update locking
Thiago Santos [Sat, 29 Nov 2014 13:28:09 +0000 (10:28 -0300)]
adaptivedemux: fix manifest update locking

To avoid unlocking a not locked mutex

9 years agoadaptivedemux: Fix segment for manifests that won't start at 0
Thiago Santos [Sat, 29 Nov 2014 13:27:25 +0000 (10:27 -0300)]
adaptivedemux: Fix segment for manifests that won't start at 0

For live streams, the manifest might start at an advanced position. Avoid
using segment start=0 to have playback start immediately

9 years agodashdemux: replace the streams with the new mpd client's streams
Thiago Santos [Sat, 29 Nov 2014 13:26:31 +0000 (10:26 -0300)]
dashdemux: replace the streams with the new mpd client's streams

Otherwise it won't update with the new segment info from the playlist
update

9 years agodashdemux: when updating the manifest, set the uri
Thiago Santos [Sat, 29 Nov 2014 13:25:45 +0000 (10:25 -0300)]
dashdemux: when updating the manifest, set the uri

This will allow dashdemux to create the correct full path URIs to
continue downloading fragments

9 years agodashdemux: port to adaptive demux
Thiago Santos [Tue, 26 Aug 2014 19:45:46 +0000 (16:45 -0300)]
dashdemux: port to adaptive demux

9 years agoadaptivedemux: also clear flushing state on pads before restarting
Thiago Santos [Sat, 22 Nov 2014 20:27:03 +0000 (17:27 -0300)]
adaptivedemux: also clear flushing state on pads before restarting

During flushing seeks the flushing flow return will propagate up to the
source element and all pads are going to have the flushing flag set.

So before restarting also remove that flag together with the EOS one.

We don't do that when pushing the flush stop event because our event
handler for the proxypad will drop all events.

9 years agoadaptivedemux: add adaptivedemux base class
Thiago Santos [Fri, 21 Nov 2014 23:42:09 +0000 (20:42 -0300)]
adaptivedemux: add adaptivedemux base class

https://bugzilla.gnome.org/show_bug.cgi?id=735848

9 years agogdppay: minor caps event payloading optimisation
Tim-Philipp Müller [Sat, 29 Nov 2014 13:16:56 +0000 (13:16 +0000)]
gdppay: minor caps event payloading optimisation

Avoid creation of buffer we're just going to throw
away two lines later anyway.

9 years agogdppay: fix some memory leaks
Tim-Philipp Müller [Sat, 29 Nov 2014 13:15:36 +0000 (13:15 +0000)]
gdppay: fix some memory leaks

9 years agoglcontext: require GL_ARB_ES2_compatibility for opengl3
Matthew Waters [Sat, 29 Nov 2014 13:42:17 +0000 (00:42 +1100)]
glcontext: require GL_ARB_ES2_compatibility for opengl3

until we generate gl3 compliant shaders

https://bugzilla.gnome.org/show_bug.cgi?id=740012

9 years agoglimagesink: properly handle the !opengl3 case
Matthew Waters [Sat, 29 Nov 2014 13:18:55 +0000 (00:18 +1100)]
glimagesink: properly handle the !opengl3 case

https://bugzilla.gnome.org/show_bug.cgi?id=740012

9 years agodecklinksink: Drop too late frames and start scheduled playback at the correct time
Sebastian Dröge [Fri, 28 Nov 2014 16:31:07 +0000 (17:31 +0100)]
decklinksink: Drop too late frames and start scheduled playback at the correct time

9 years agoglfilterapp: fix unused variable compiler warning
Tim-Philipp Müller [Fri, 28 Nov 2014 16:04:54 +0000 (16:04 +0000)]
glfilterapp: fix unused variable compiler warning

When GST_GL_HAVE_OPENGL is 0 or unset.

9 years agodecklinksrc: For the source, unsetting the callback already deletes it
Sebastian Dröge [Fri, 28 Nov 2014 14:06:42 +0000 (15:06 +0100)]
decklinksrc: For the source, unsetting the callback already deletes it

9 years agodecklinksink: Wait for the pipeline clock instead of the decklink clock
Sebastian Dröge [Fri, 14 Nov 2014 10:19:45 +0000 (11:19 +0100)]
decklinksink: Wait for the pipeline clock instead of the decklink clock

Otherwise we're going to starve other elements if the decklink clock
is slower than the pipeline clock, or starts much later.

Of course this will still cause problems if the decklink clock and ours are
completely out of sync, or running at a very different rate. But this at least
works better now.

9 years agodecklinksink: Schedule video frames according to their timestamps, not according...
Sebastian Dröge [Fri, 14 Nov 2014 09:58:53 +0000 (10:58 +0100)]
decklinksink: Schedule video frames according to their timestamps, not according to a frame counter

9 years agodecklinksink: Add some FIXME comments about synchronization
Sebastian Dröge [Fri, 14 Nov 2014 10:02:01 +0000 (11:02 +0100)]
decklinksink: Add some FIXME comments about synchronization

9 years agodecklinksrc: Sample the pipeline clock for the timestamps instead of coming up with...
Sebastian Dröge [Fri, 14 Nov 2014 09:38:33 +0000 (10:38 +0100)]
decklinksrc: Sample the pipeline clock for the timestamps instead of coming up with our own

If we just count the frames and calculate timestamps from that, all frames
will arrive late in the sink as we have a live source here. Instead take
the pipeline clock at capture time as reference.

9 years agodecklinksrc: Handle callback/delegate object the same way as in the sink
Sebastian Dröge [Fri, 28 Nov 2014 13:50:18 +0000 (14:50 +0100)]
decklinksrc: Handle callback/delegate object the same way as in the sink

Potentially fixes crashes.

9 years agodecklinksink: Don't crash when going to READY another time
Sebastian Dröge [Fri, 28 Nov 2014 13:46:06 +0000 (14:46 +0100)]
decklinksink: Don't crash when going to READY another time

We have to handle the callback object a bit different:
a) it needs a virtual destructor
b) we need to set the callback to NULL when we're done with the output
c) create a new one every time

https://bugzilla.gnome.org/show_bug.cgi?id=740616

9 years agocompositor: Failure to map a video frame is not just a warning
Sebastian Dröge [Fri, 28 Nov 2014 09:23:55 +0000 (10:23 +0100)]
compositor: Failure to map a video frame is not just a warning

Also add some warning debug output if mapping a output buffer fails

9 years agovideoaggregator: Failure to map a video frame is not just a warning
Sebastian Dröge [Fri, 28 Nov 2014 09:22:44 +0000 (10:22 +0100)]
videoaggregator: Failure to map a video frame is not just a warning

9 years agoglfilterapp is now available on gles2 as well so build it
Matthew Waters [Fri, 28 Nov 2014 03:08:40 +0000 (14:08 +1100)]
glfilterapp is now available on gles2 as well so build it

9 years agogl: fixup compat definition for GLuint64 for OS X
Matthew Waters [Fri, 28 Nov 2014 02:04:21 +0000 (13:04 +1100)]
gl: fixup compat definition for GLuint64 for OS X

9 years agogl: add compat definition for GLuint64 for android
Matthew Waters [Fri, 28 Nov 2014 00:56:10 +0000 (11:56 +1100)]
gl: add compat definition for GLuint64 for android

../../../../gst-libs/gst/gl/glprototypes/sync.h:41:23: error: unknown type name 'GLuint64'
                       GLuint64 timeout))

9 years agobasecamerasrc: fix typo in docs
Danny Song [Thu, 27 Nov 2014 17:21:49 +0000 (02:21 +0900)]
basecamerasrc: fix typo in docs

https://bugzilla.gnome.org/show_bug.cgi?id=740815

9 years agoglsync: fix build with desktop gl
Matthew Waters [Fri, 28 Nov 2014 00:11:43 +0000 (11:11 +1100)]
glsync: fix build with desktop gl

9 years agogl: fix typo in #if
Matthew Waters [Fri, 28 Nov 2014 00:04:22 +0000 (11:04 +1100)]
gl: fix typo in #if

GST_GL_API_HAVE_OPENGL doesn't exist

9 years agogl: fix various build errors without desktop gl
Matthew Waters [Thu, 27 Nov 2014 23:59:46 +0000 (10:59 +1100)]
gl: fix various build errors without desktop gl

9 years agogldisplay: fix build error
Matthew Waters [Thu, 27 Nov 2014 23:03:18 +0000 (10:03 +1100)]
gldisplay: fix build error

gstgldisplay.c:234:541: error: 'return' with no value, in function returning non-void [-Werror]
   g_return_if_fail (GST_IS_GL_DISPLAY (display));

9 years agogldisplay: implement runtime GL api filtering
Matthew Waters [Thu, 27 Nov 2014 10:05:45 +0000 (21:05 +1100)]
gldisplay: implement runtime GL api filtering

Needed so that the pipeline/application can limit the choice of GL api
to what it supports

9 years agogl: remove the use of glu
Matthew Waters [Thu, 27 Nov 2014 05:17:50 +0000 (16:17 +1100)]
gl: remove the use of glu

9 years agogl: add a sync meta for synchronizing across GL contexts
Matthew Waters [Fri, 17 Oct 2014 13:22:24 +0000 (15:22 +0200)]
gl: add a sync meta for synchronizing across GL contexts

A context can create a GLsync object that can be waited on in order
to ensure that GL resources created in one context are able to be
used in another shared context without any chance of reading invalid
data.

This meta would be placed on buffers that are known to cross from
one context to another.  The receiving element would then wait
on the sync object to ensure that the data to be used is complete.

9 years agoglprototypes: add sync function definitions
Matthew Waters [Fri, 17 Oct 2014 13:08:29 +0000 (15:08 +0200)]
glprototypes: add sync function definitions

9 years agoglfilterapp: port to gles2 and gl3
Matthew Waters [Mon, 17 Nov 2014 07:06:20 +0000 (18:06 +1100)]
glfilterapp: port to gles2 and gl3

9 years agogloverlay: support gl3
Matthew Waters [Fri, 14 Nov 2014 12:24:48 +0000 (23:24 +1100)]
gloverlay: support gl3

9 years agoglvideomixer: add support for gl3
Matthew Waters [Thu, 13 Nov 2014 13:20:10 +0000 (00:20 +1100)]
glvideomixer: add support for gl3

9 years agoglfiltercube: add a gl3 code path
Matthew Waters [Thu, 13 Nov 2014 10:30:38 +0000 (21:30 +1100)]
glfiltercube: add a gl3 code path

9 years agoglfiltershader: support gl3
Matthew Waters [Thu, 13 Nov 2014 04:09:04 +0000 (15:09 +1100)]
glfiltershader: support gl3

9 years agogltransformation: support gl3
Matthew Waters [Thu, 13 Nov 2014 01:12:46 +0000 (12:12 +1100)]
gltransformation: support gl3

9 years agogleffects: support gl3 whereever gles2 is supported
Matthew Waters [Wed, 12 Nov 2014 12:48:32 +0000 (23:48 +1100)]
gleffects: support gl3 whereever gles2 is supported

9 years agoglcolorscale: support gl3
Matthew Waters [Wed, 12 Nov 2014 12:47:13 +0000 (23:47 +1100)]
glcolorscale: support gl3

9 years agoglfilter: add a gl3 code path using vao's and vbo's
Matthew Waters [Wed, 12 Nov 2014 11:49:11 +0000 (22:49 +1100)]
glfilter: add a gl3 code path using vao's and vbo's

9 years agoglframebuffer: add support for gl3
Matthew Waters [Wed, 12 Nov 2014 11:48:04 +0000 (22:48 +1100)]
glframebuffer: add support for gl3

9 years agoglcolorconvert: add support for gl3
Matthew Waters [Thu, 23 Oct 2014 11:35:20 +0000 (22:35 +1100)]
glcolorconvert: add support for gl3

9 years agoglimagesink: use vbo's and vao's for gl3
Matthew Waters [Tue, 21 Oct 2014 11:18:51 +0000 (22:18 +1100)]
glimagesink: use vbo's and vao's for gl3

9 years agoglshader add support for gl3
Matthew Waters [Tue, 21 Oct 2014 08:30:38 +0000 (19:30 +1100)]
glshader add support for gl3

9 years agoglx: ask for a GL3 core context
Matthew Waters [Sun, 19 Oct 2014 04:09:37 +0000 (08:09 +0400)]
glx: ask for a GL3 core context

9 years agovideoaggregator: Minor cleanup
Sebastian Dröge [Thu, 27 Nov 2014 20:22:44 +0000 (21:22 +0100)]
videoaggregator: Minor cleanup

9 years agocompositor: Implement rescaling of the input via pad properties
Sebastian Dröge [Wed, 26 Nov 2014 14:02:14 +0000 (15:02 +0100)]
compositor: Implement rescaling of the input via pad properties

compositor has now the same interface as glvideomixer.

9 years agovideoaggregator: Copy over more fields from the relevant video-info
Sebastian Dröge [Thu, 27 Nov 2014 19:34:25 +0000 (20:34 +0100)]
videoaggregator: Copy over more fields from the relevant video-info

gst_video_info_set_format() will reset the complete video-info, but
we want to keep values like the PAR, colorimetry and chroma site.
Otherwise we risk setting different values on the srcpad caps than
what is actually inside the buffers.

9 years agovideoaggregator: Directly use the converters video-info instead of recalculating it
Sebastian Dröge [Thu, 27 Nov 2014 19:25:29 +0000 (20:25 +0100)]
videoaggregator: Directly use the converters video-info instead of recalculating it

9 years agovideoaggregator: Do source pad negotiation only from the aggregated function
Sebastian Dröge [Thu, 27 Nov 2014 18:52:20 +0000 (19:52 +0100)]
videoaggregator: Do source pad negotiation only from the aggregated function

Otherwise we might negotiate from the sinkpad streaming threads at
the same time as on the srcpad streaming thread, and then all kinds
of crazy bugs happen that don't make any sense at all.

9 years agovideoconvert: Hide all conversion related fields
Thibault Saunier [Thu, 27 Nov 2014 17:46:03 +0000 (18:46 +0100)]
videoconvert: Hide all conversion related fields

And do not delay the setting of the conversion_info

https://bugzilla.gnome.org/show_bug.cgi?id=740768

9 years agoaudiomixer: Do not try to resize a buffer to a negative size on EOS
Thibault Saunier [Thu, 27 Nov 2014 15:43:39 +0000 (16:43 +0100)]
audiomixer: Do not try to resize a buffer to a negative size on EOS

9 years agovideoaggregator: Expose vmethods to set converters and prepare/clean frames
Thibault Saunier [Wed, 26 Nov 2014 17:24:05 +0000 (18:24 +0100)]
videoaggregator: Expose vmethods to set converters and prepare/clean frames

This gives more flexibility to the subclasses and permits to remove the
GstVideoAggregatorClass->disable_frame_conversion ugly API.

WARNING: This breaks the API as it removes the disable_frame_conversion
field

API:
  + GstVideoAggregatorClass->find_best_format
  + GstVideoAggregatorPadClass->set_format
  + GstVideoAggregatorPadClass->prepare_frame
  + GstVideoAggregatorPadClass->clean_frame

  - GstVideoAggregatorClass->disable_frame_conversion

https://bugzilla.gnome.org/show_bug.cgi?id=740768

9 years agowatchdog: Handle FLUSHING seek events even in PAUSED state
Thibault Saunier [Wed, 26 Nov 2014 14:41:37 +0000 (15:41 +0100)]
watchdog: Handle FLUSHING seek events even in PAUSED state

https://bugzilla.gnome.org/show_bug.cgi?id=740757

9 years agowatchdog: Properly handle PAUSED state
Thibault Saunier [Wed, 26 Nov 2014 14:43:12 +0000 (15:43 +0100)]
watchdog: Properly handle PAUSED state

In PAUSED state we should just preroll, so get one buffer
and stop monitoring.

https://bugzilla.gnome.org/show_bug.cgi?id=740757

9 years agovideoaggregator: Let a full renegotiation happen after removing the last pad
Thibault Saunier [Fri, 31 Oct 2014 10:01:47 +0000 (11:01 +0100)]
videoaggregator: Let a full renegotiation happen after removing the last pad

With the current code, we will end up setting the preferred downstream
format as the srcpad format, and it might not be accepted by the next
sinkpad to be added. We should instead let the next sinkpad reconfigure
everything.

9 years agovtenc: add quality property
Alessandro Decina [Thu, 27 Nov 2014 12:40:56 +0000 (13:40 +0100)]
vtenc: add quality property

9 years agovtenc: set profile and level in caps
Alessandro Decina [Thu, 27 Nov 2014 12:33:07 +0000 (13:33 +0100)]
vtenc: set profile and level in caps

9 years agoglcolorconvert: Unref buffer with the correct function
Sebastian Dröge [Thu, 27 Nov 2014 17:09:58 +0000 (18:09 +0100)]
glcolorconvert: Unref buffer with the correct function

9 years agoglmemory: Handle failure of memory allocation gracefully
Sebastian Dröge [Thu, 27 Nov 2014 17:09:14 +0000 (18:09 +0100)]
glmemory: Handle failure of memory allocation gracefully

9 years agoglmemory: Use g_try_malloc() in the appropriate places
Sebastian Dröge [Thu, 27 Nov 2014 17:04:58 +0000 (18:04 +0100)]
glmemory: Use g_try_malloc() in the appropriate places

g_malloc() aborts if allocation fails, it's pointless to check against NULL
afterwards. That's why g_try_malloc() exists.