platform/upstream/gstreamer.git
8 years agogaudi: fix documentation warning in GstChromium
Luis de Bethencourt [Thu, 7 Jul 2016 16:16:47 +0000 (17:16 +0100)]
gaudi: fix documentation warning in GstChromium

Fixing the following warning when generating documentation:
xml/element-chromium.xml:74: element refsect2: validity error :
    ID GstChromium already defined
<refsect2 id="GstChromium" role="typedef">
                                         ^
Warning: multiple "IDs" for constraint linkend: GstChromium.
  DOC   Fixing cross-references

8 years agowebrtcdsp: Relax C++11 check
Nicolas Dufresne [Thu, 7 Jul 2016 16:31:20 +0000 (12:31 -0400)]
webrtcdsp: Relax C++11 check

The library does not require any STL C++11 specific headers.
This fixes our OSX build as we target a SDK version that does not
support C++11 STL.

8 years agowebrtcdsp: Add Android GNUSTL support
Nicolas Dufresne [Thu, 7 Jul 2016 16:31:03 +0000 (12:31 -0400)]
webrtcdsp: Add Android GNUSTL support

8 years agowebrtcdsp: Set USE_WEBRTCDSP when disabled
Nicolas Dufresne [Thu, 7 Jul 2016 16:20:45 +0000 (12:20 -0400)]
webrtcdsp: Set USE_WEBRTCDSP when disabled

8 years agoadaptivedemux: Also consider more seek flags as "snap" seeks
Edward Hervey [Thu, 7 Jul 2016 15:20:58 +0000 (17:20 +0200)]
adaptivedemux: Also consider more seek flags as "snap" seeks

In the case of KEY_UNIT and TRICKMODE_KEY_UNITS seeks, we want to
"snap" to the closest fragment.

Without this, we end up pushing out a segment which does not match
the first fragment timestamp being pushed out, resulting in one or
more buffers being eventually dropped because they are out of segment.

8 years agowebrtcdsp: Also allow webrtc-audio-processing 0.3
Sebastian Dröge [Wed, 6 Jul 2016 11:29:04 +0000 (14:29 +0300)]
webrtcdsp: Also allow webrtc-audio-processing 0.3

It seems to work and is the version available in Debian.

8 years agomusepackdeck: unsigned can't be less than zero
Luis de Bethencourt [Wed, 6 Jul 2016 11:12:41 +0000 (12:12 +0100)]
musepackdeck: unsigned can't be less than zero

Less-than-zero comparison of an unsigned value, guint64, is never true

CID 1363334

8 years agoBack to development
Sebastian Dröge [Wed, 6 Jul 2016 10:51:06 +0000 (13:51 +0300)]
Back to development

8 years agoRelease 1.9.1
Sebastian Dröge [Wed, 6 Jul 2016 10:07:21 +0000 (13:07 +0300)]
Release 1.9.1

8 years agoUpdate .po files
Sebastian Dröge [Wed, 6 Jul 2016 08:52:11 +0000 (11:52 +0300)]
Update .po files

8 years agopo: Update translations
Sebastian Dröge [Wed, 6 Jul 2016 07:19:24 +0000 (10:19 +0300)]
po: Update translations

8 years agowebrtcdsp: Workaround GCC/MingW 4.7 limitations
Nicolas Dufresne [Tue, 5 Jul 2016 13:21:58 +0000 (09:21 -0400)]
webrtcdsp: Workaround GCC/MingW 4.7 limitations

It refuses to initialize a classes using brackets notation. This is
to allow building using our mingw version.

8 years agoconfigure.ac: Add missing double quotes
Nicolas Dufresne [Mon, 4 Jul 2016 18:57:11 +0000 (14:57 -0400)]
configure.ac: Add missing double quotes

That affectation was ignored. This worked on recent GCC as C++11 is
enabled by default now.

8 years agolv2: add support for saving presets
Stefan Sauer [Sat, 2 Jul 2016 08:24:51 +0000 (10:24 +0200)]
lv2: add support for saving presets

Plugns supporting the state interface can now save their presets under '.lv2'.

8 years agolv2: support for loading presets
Stefan Sauer [Wed, 15 Jun 2016 20:22:58 +0000 (22:22 +0200)]
lv2: support for loading presets

Detect if plugins can do presets. Lazily read a list of presets and add support
for loading.

8 years agomusepackdec: port to 1.0
Wim Taymans [Mon, 4 Jul 2016 14:54:53 +0000 (16:54 +0200)]
musepackdec: port to 1.0

8 years agoshmsrc: Do not call gst_poll_remove_fd upon stop if reading was never started
Marcin Lewandowski [Mon, 4 Jul 2016 10:19:36 +0000 (12:19 +0200)]
shmsrc: Do not call gst_poll_remove_fd upon stop if reading was never started

https://bugzilla.gnome.org/show_bug.cgi?id=768365

8 years agodvb/parsechannels: improve parsing-success message
Reynaldo H. Verdejo Pinochet [Fri, 1 Jul 2016 21:50:27 +0000 (14:50 -0700)]
dvb/parsechannels: improve parsing-success message

* Fix typo
* Give information on what was actually parsed

8 years agodvb/parsechannels: don't use default conf filename for messages
Reynaldo H. Verdejo Pinochet [Fri, 1 Jul 2016 21:37:37 +0000 (14:37 -0700)]
dvb/parsechannels: don't use default conf filename for messages

The file name can be user-defined.

8 years agoqt: fix build some more when QPA is not available
Tim-Philipp Müller [Fri, 1 Jul 2016 18:22:32 +0000 (19:22 +0100)]
qt: fix build some more when QPA is not available

Compiler would complain about include directory that didn't
exist because QPA_INCLUDE_PATH gets subst-ed regardless
(and if it didn't we'd have just an empty -I argument).

https://bugzilla.gnome.org/show_bug.cgi?id=767553

8 years agogstjpeg2000parse: no need for else after a goto
Luis de Bethencourt [Fri, 1 Jul 2016 13:05:49 +0000 (14:05 +0100)]
gstjpeg2000parse: no need for else after a goto

else clause is redundant when the end of the if statement is a jump.
If we haven't gone to the beach, we are in the false branch.

8 years agogstjpeg2000parse: improved parsing of jpc magic and j2c box
Aaron Boxer [Thu, 30 Jun 2016 11:47:50 +0000 (07:47 -0400)]
gstjpeg2000parse: improved parsing of jpc magic and j2c box

https://bugzilla.gnome.org/show_bug.cgi?id=767512

8 years agoadaptivedemux: Add custom flow return for allowing subclasses to specify when a fragm...
Sebastian Dröge [Wed, 25 May 2016 13:01:24 +0000 (16:01 +0300)]
adaptivedemux: Add custom flow return for allowing subclasses to specify when a fragment is finished

If it is finished before upstream going EOS.

https://bugzilla.gnome.org/show_bug.cgi?id=767365

8 years agodashdemux: Implement SIDX tracking based on buffer offset
Sebastian Dröge [Fri, 27 May 2016 13:08:09 +0000 (16:08 +0300)]
dashdemux: Implement SIDX tracking based on buffer offset

This simplifies the code but also removes a bug with tracking of the remaining
size for the initial subfragment: we were not considering the size between the
index and the start of the first moof here.

https://bugzilla.gnome.org/show_bug.cgi?id=764684

8 years agodashdemux: Properly keep track of current offset
Sebastian Dröge [Thu, 31 Mar 2016 10:39:59 +0000 (13:39 +0300)]
dashdemux: Properly keep track of current offset

GstAdapter does not guarantee to pass through all the offsets, we have to keep
track of it ourselves.

https://bugzilla.gnome.org/show_bug.cgi?id=764684

8 years agohlsdemux: Properly keep track of current offset
Sebastian Dröge [Thu, 31 Mar 2016 10:39:59 +0000 (13:39 +0300)]
hlsdemux: Properly keep track of current offset

GstAdapter does not guarantee to pass through all the offsets, we have to keep
track of it ourselves.

https://bugzilla.gnome.org/show_bug.cgi?id=764684

8 years agohlsdemux: Clear pending data when needed
Sebastian Dröge [Thu, 31 Mar 2016 10:01:00 +0000 (13:01 +0300)]
hlsdemux: Clear pending data when needed

When switching fragments we don't want to keep any data around from the last
one, and also forget about all data when doing flushing seeks or selecting new
bitrates.

https://bugzilla.gnome.org/show_bug.cgi?id=764684

8 years agoadaptivedemux: Get rid of internal stream adapter and let subclasses handle this...
Sebastian Dröge [Mon, 28 Mar 2016 10:45:36 +0000 (13:45 +0300)]
adaptivedemux: Get rid of internal stream adapter and let subclasses handle this directly

This allows subclasses to have more control and especially ensure that they
push data downstream with the correct offsets.

https://bugzilla.gnome.org/show_bug.cgi?id=764684

8 years agocodecparsers: jpeg2000: don't allow 0 value for array access with i-1
Tim-Philipp Müller [Fri, 1 Jul 2016 11:20:14 +0000 (12:20 +0100)]
codecparsers: jpeg2000: don't allow 0 value for array access with i-1

8 years agocodecparsers: jpeg2000: fix possible out-of-bounds array access
Tim-Philipp Müller [Fri, 1 Jul 2016 11:15:28 +0000 (12:15 +0100)]
codecparsers: jpeg2000: fix possible out-of-bounds array access

sizeof(array) != G_N_ELEMENTS(array).

CID 1362900

8 years agoopenh264enc: Set frame timestamps before sending to the encoder
Sebastian Dröge [Thu, 30 Jun 2016 21:38:26 +0000 (23:38 +0200)]
openh264enc: Set frame timestamps before sending to the encoder

8 years agoopenh264enc: Fix initial time-per-frame calculation
Sebastian Dröge [Thu, 30 Jun 2016 21:35:33 +0000 (23:35 +0200)]
openh264enc: Fix initial time-per-frame calculation

8 years agoopenh264enc: Remove meaningless drop bitrate handling
Sebastian Dröge [Thu, 30 Jun 2016 21:33:38 +0000 (23:33 +0200)]
openh264enc: Remove meaningless drop bitrate handling

This doesn't even have a property.

8 years agoopenh264enc: Expose maximum bitrate setting
Sebastian Dröge [Thu, 30 Jun 2016 21:29:13 +0000 (23:29 +0200)]
openh264enc: Expose maximum bitrate setting

8 years agoopenh264enc: Actually hook up the rate-control property
Sebastian Dröge [Thu, 30 Jun 2016 21:19:26 +0000 (23:19 +0200)]
openh264enc: Actually hook up the rate-control property

8 years agoopenh264enc: Use a constant SPS/PPS ID no matter if openh264 older or newer than...
Sebastian Dröge [Thu, 30 Jun 2016 21:06:29 +0000 (23:06 +0200)]
openh264enc: Use a constant SPS/PPS ID no matter if openh264 older or newer than 1.4 is used

8 years agoopenh264enc: Make slice settings more explicit and don't set any number if not a...
Sebastian Dröge [Thu, 30 Jun 2016 21:02:49 +0000 (23:02 +0200)]
openh264enc: Make slice settings more explicit and don't set any number if not a fixed number of slices is selected

8 years agowebrtcdsp: Rewrite echo data synchronization
Nicolas Dufresne [Wed, 29 Jun 2016 19:56:47 +0000 (15:56 -0400)]
webrtcdsp: Rewrite echo data synchronization

The previous code would run out of sync if there was packet lost
or clock skews. When that happened, the echo cancellation feature would
completely stop working. As this is crucial for audio calls, this patch
re-implement synchronization completely.

Instead of letting it drift until next discont, we now synchronize
against the record data at every iteration. This way we simply never
let the stream drift for longer then 10ms period. We also shorter the
delay by using the latency up the probe (basically excluding the sink
latency. This is a decent delay to avoid starving in the probe queue.

https://bugzilla.gnome.org/show_bug.cgi?id=768009

8 years agowebrtcdsp: We now fail if there is no echo probe
Nicolas Dufresne [Wed, 29 Jun 2016 19:55:19 +0000 (15:55 -0400)]
webrtcdsp: We now fail if there is no echo probe

When echo cancel is enabled, we now fail the pipeline if there is
not echo probe. For this reason there is no need to check if probe
pointer is set anymore.

8 years agoglviewconvert: fix transformation matrix usage
Matthew Waters [Thu, 30 Jun 2016 11:50:12 +0000 (21:50 +1000)]
glviewconvert: fix transformation matrix usage

Calling glUniformMatrix before the shader is bound is invalid and
would result in errors like:

GL_INVALID_OPERATION in glUniformMatrix(program not linked)

Move glUniformMatrix() to after the gst_gl_shader_use() call.

8 years agokmssink: fallback to universal planes if no overlay plane is found
Javier Martinez Canillas [Tue, 28 Jun 2016 20:07:43 +0000 (16:07 -0400)]
kmssink: fallback to universal planes if no overlay plane is found

Without setting the DRM_CLIENT_CAP_UNIVERSAL_PLANES capability bit, only
overlay planes are made available for compatibility with legacy clients.

But if a CRTC doesn't have an overlay plane associated, then kmssink is
not able to find a plane for the CRTC and the pipeline will fail, i.e:

ERROR                kmssink gstkmssink.c:482:gst_kms_sink_start:<kmssink0> Could not find a plane for crtc

If no overlay planes were found for a given CRTC, fallback to universal
planes so DRM will also return primary planes that can be used instead.

https://bugzilla.gnome.org/show_bug.cgi?id=768183

Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
8 years agokmssink: add a plane-id property
Javier Martinez Canillas [Wed, 29 Jun 2016 00:51:56 +0000 (20:51 -0400)]
kmssink: add a plane-id property

Without setting the DRM_CLIENT_CAP_UNIVERSAL_PLANES capability bit, only
overlay planes are made available for compatibility with legacy clients.

But if a CRTC doesn't have an overlay plane associated, then kmssink is
not able to find a plane for the CRTC and the pipeline will fail, i.e:

ERROR                kmssink gstkmssink.c:482:gst_kms_sink_start:<kmssink0> Could not find a plane for crtc

This patch adds a plane-id property to the kmssink element so a specific
plane can be used in case that a CRTC has only a primary plane associated.

https://bugzilla.gnome.org/show_bug.cgi?id=768183

8 years agoopenh264enc: Remove broken byte-stream to avc conversion and just output byte-stream...
Sebastian Dröge [Thu, 30 Jun 2016 08:33:30 +0000 (10:33 +0200)]
openh264enc: Remove broken byte-stream to avc conversion and just output byte-stream as generated by the encoder

The byte-stream to avc conversion did not consider NAL sizes bigger than 2^16,
multiple layers, multiple NALs per layer, and various other things. This
caused corrupted streams in higher bitrates and other circumstances.

Let's just forward byte-stream as generated by the encoder and let h264parse
handle conversion to avc if needed. That way we only have to keep around one
version of the conversion and don't have to fix it in multiple places.

8 years agoh264parse: Drop filler data
Edward Hervey [Wed, 29 Jun 2016 16:36:56 +0000 (18:36 +0200)]
h264parse: Drop filler data

When skipping data, check if they are filler bytes. If so, drop the
data instead of skipping. We don't want to output filler bytes, but they
shouldn't cause a discontinuity.

https://bugzilla.gnome.org/show_bug.cgi?id=768125

8 years agodvbbasebin: Use a non-timer GstPoll
Sebastian Dröge [Thu, 30 Jun 2016 06:31:20 +0000 (08:31 +0200)]
dvbbasebin: Use a non-timer GstPoll

set_flushing() causes inconsistent states on timer polls and also signals a
g_critical() nowadays because of that.

8 years agoglupload: fix raw data uploader with addition of GstVideoGLTextureType
Matthew Waters [Thu, 30 Jun 2016 04:04:36 +0000 (14:04 +1000)]
glupload: fix raw data uploader with addition of GstVideoGLTextureType

Fixes regression from 989200820d43c78fb179d7984827800d83787390

https://bugzilla.gnome.org/show_bug.cgi?id=768217

8 years agoglvideoconvert: RG/LA is available in OpenGL ES 3, so let that through
Olivier Crête [Thu, 30 Jun 2016 03:55:18 +0000 (23:55 -0400)]
glvideoconvert: RG/LA is available in OpenGL ES 3, so let that through

8 years agovtdec: use L/LA textures on GLES2 as well
Matthew Waters [Wed, 29 Jun 2016 08:35:17 +0000 (18:35 +1000)]
vtdec: use L/LA textures on GLES2 as well

8 years agoRevert "Revert "gl/eagl: try getting a gles3 context""
Matthew Waters [Fri, 24 Jun 2016 03:41:11 +0000 (13:41 +1000)]
Revert "Revert "gl/eagl: try getting a gles3 context""

This reverts commit eb142736318463e6e553c21707ba0d716f85a46f.

vtdec now successfully outputs textures that can be converted by
glcolorconvert in GLES3 contexts.

8 years agovtdec: add support for zerocopy decoding with GLES 3.x
Matthew Waters [Fri, 24 Jun 2016 14:05:37 +0000 (00:05 +1000)]
vtdec: add support for zerocopy decoding with GLES 3.x

It seems that Red/RG textures aren't supported anymore for zerocopy decoding
with GLES 3.x.  Instead use Luminance/Luminance Alpha textures which are.

8 years agoglmemory: add the texture type to allocate to parameters
Matthew Waters [Tue, 28 Jun 2016 03:51:22 +0000 (13:51 +1000)]
glmemory: add the texture type to allocate to parameters

Rather than assuming something.  e.g. zerocopy on iOS with GLES3 requires
the use of Luminance/Luminance Alpha formats and does not work with
Red/RG textures.

8 years agoglcolorconvert: don't assume Red/RG textures will be used
Matthew Waters [Tue, 28 Jun 2016 03:09:51 +0000 (13:09 +1000)]
glcolorconvert: don't assume Red/RG textures will be used

Take the used texture type from the memory instead.

Fixes conversion from multi-planar YUV formats with two components per plane
(NV12, NV21, YUY2, UYVY, GRAY16_*, etc) with Luminance Alpha input textures.

This is also needed for zerocopy decoding on iOS with GLES 3.x.

8 years agoglcontext: correct precondition for get_gl_version
Matthew Waters [Tue, 28 Jun 2016 03:02:32 +0000 (13:02 +1000)]
glcontext: correct precondition for get_gl_version

The intention was to assert if both maj and min were NULL (as there would be no
point calling the function).  Instead if either maj or min were NULL, the assert
would occur.

Fix that.

8 years agod3dvideosink: Replace some more gst_caps_to_string() with GST_PTR_FORMAT
Sebastian Dröge [Tue, 28 Jun 2016 16:26:29 +0000 (19:26 +0300)]
d3dvideosink: Replace some more gst_caps_to_string() with GST_PTR_FORMAT

https://bugzilla.gnome.org/show_bug.cgi?id=768146

8 years agod3dvideosink: leak on caps string
Jonathan Roy [Tue, 28 Jun 2016 15:50:59 +0000 (11:50 -0400)]
d3dvideosink: leak on caps string

https://bugzilla.gnome.org/show_bug.cgi?id=768146

8 years agojpeg2000parse: fix typo in sampling check
Luis de Bethencourt [Tue, 28 Jun 2016 10:08:22 +0000 (11:08 +0100)]
jpeg2000parse: fix typo in sampling check

Fixing small typo that changes the value of the return.

8 years agotests: Ignore kmssink for state change test
Olivier Crête [Tue, 28 Jun 2016 01:36:50 +0000 (21:36 -0400)]
tests: Ignore kmssink for state change test

It can only be put to READY if the DRM stuff is available.

8 years agoqmlglsink: Fix build error when don't have QPA installed.
Haihua Hu [Mon, 27 Jun 2016 10:15:08 +0000 (18:15 +0800)]
qmlglsink: Fix build error when don't have QPA installed.

Check header file existance and wrap the header file include
in the necessary #ifdef to avoid build error.

https://bugzilla.gnome.org/show_bug.cgi?id=767553

8 years agodvbbasebin: make sure we have an error in case of parsing failure
Reynaldo H. Verdejo Pinochet [Sat, 25 Jun 2016 01:14:06 +0000 (18:14 -0700)]
dvbbasebin: make sure we have an error in case of parsing failure

Drop a redundant comment and rellocate another one while at it.

8 years agodvb/parsechannels: implement channels.conf format autodetection
Reynaldo H. Verdejo Pinochet [Wed, 22 Jun 2016 23:14:33 +0000 (16:14 -0700)]
dvb/parsechannels: implement channels.conf format autodetection

Simple autodetection avoids having to call the parsers
one after the other.

8 years agodvb/parsechannels: rename v5/zap parsers for congruency
Reynaldo H. Verdejo Pinochet [Tue, 21 Jun 2016 23:08:57 +0000 (16:08 -0700)]
dvb/parsechannels: rename v5/zap parsers for congruency

Add function prototypes while at it.

8 years agodvb/parsechannels: fix modulation names
Reynaldo H. Verdejo Pinochet [Wed, 13 Apr 2016 04:04:25 +0000 (21:04 -0700)]
dvb/parsechannels: fix modulation names

Some names were incorrect. Authoritative source for
the dvbv5 format taken from v4l-utils' lib/libdvbv5/dvb-v5.c

Aditionally, add the missing setter mapping for the
modulation param.

This change makes ATSC work.

https://bugzilla.gnome.org/show_bug.cgi?id=764957

8 years agodvbbasebin: also accept DVBv5 channels.conf format
Thiago Santos [Thu, 25 Feb 2016 00:44:40 +0000 (21:44 -0300)]
dvbbasebin: also accept DVBv5 channels.conf format

Confirmed working with ISDB-T & ATSC conf files.

https://bugzilla.gnome.org/show_bug.cgi?id=764957

8 years agodvbsrc: fix ISDB-T property descriptions
Reynaldo H. Verdejo Pinochet [Tue, 21 Jun 2016 18:48:36 +0000 (11:48 -0700)]
dvbsrc: fix ISDB-T property descriptions

Drop some stale trailing whitespace while at it.

8 years agodvbsrc: fix misplaced space on doc block
Reynaldo H. Verdejo Pinochet [Tue, 21 Jun 2016 18:32:07 +0000 (11:32 -0700)]
dvbsrc: fix misplaced space on doc block

8 years agoh264parse: When input alignment=AU, always drain each packet.
Jan Schmidt [Mon, 9 May 2016 17:32:59 +0000 (03:32 +1000)]
h264parse: When input alignment=AU, always drain each packet.

If the input alignment claims AU alignment, each received
buffer should contain a complete video frame, so never hold over parts
of buffers for later processing. Also reduces latency, as packets
are parsed/converted and output immediately instead of 1 buffer
later.

Fixes a problem where an (arguably disallowed) padding byte on the
end of a buffer is detected as an extra byte in the following
start code, and messes up the timestamping that should apply to
that start code.

8 years agowin32: update codecparsers .def file for new symbols
Tim-Philipp Müller [Fri, 24 Jun 2016 12:05:24 +0000 (13:05 +0100)]
win32: update codecparsers .def file for new symbols

8 years agoopenjpeg: fix more broken includes
Tim-Philipp Müller [Fri, 24 Jun 2016 08:41:18 +0000 (09:41 +0100)]
openjpeg: fix more broken includes

8 years agovideoparsers: fix broken include
Tim-Philipp Müller [Fri, 24 Jun 2016 08:39:00 +0000 (09:39 +0100)]
videoparsers: fix broken include

8 years agojpeg2000parse: use enums for colorspace and sampling, rather than strings
Aaron Boxer [Tue, 21 Jun 2016 16:41:46 +0000 (12:41 -0400)]
jpeg2000parse: use enums for colorspace and sampling, rather than strings

Also, move gstjpeg2000sampling to codecparsers project

https://bugzilla.gnome.org/show_bug.cgi?id=767908

8 years agoconfigure: Need to add -DGST_STATIC_COMPILATION when building only statically
Nirbheek Chauhan [Thu, 23 Jun 2016 20:27:13 +0000 (01:57 +0530)]
configure: Need to add -DGST_STATIC_COMPILATION when building only statically

https://bugzilla.gnome.org/show_bug.cgi?id=767463

8 years agoeglimage: Only register debug category if dmabuf support is enabled
Sebastian Dröge [Thu, 23 Jun 2016 12:57:58 +0000 (15:57 +0300)]
eglimage: Only register debug category if dmabuf support is enabled

It's not used otherwise currently and causes compiler warnings.

8 years agowebrtcdsp: Offset timestamp with duration
Nicolas Dufresne [Thu, 23 Jun 2016 02:28:03 +0000 (22:28 -0400)]
webrtcdsp: Offset timestamp with duration

The saved timestamp is used to compute the delay of the probe data.
As it's used at the following incoming buffer, it needs to be offset
with the duration of the buffer to represent the end position. Also,
properly initialize the saved timestamp and protect against TIME_NONE.

8 years agowebrtcdsp: Synchronize with delays
Nicolas Dufresne [Thu, 23 Jun 2016 01:54:13 +0000 (21:54 -0400)]
webrtcdsp: Synchronize with delays

Until now, we were synchronizing both DSP and Probe adapter by
waiting and clipping the probe adapter data. This increases the CPU
usage, can cause copies if the audio is not 10ms aligned and the worst
is that it prevents the processing from compensating for inaccurate
latency. This is also a step forward toward supporting playback
filters.

8 years agowebrtdsp: Remove restriction on channels number
Nicolas Dufresne [Thu, 23 Jun 2016 01:45:08 +0000 (21:45 -0400)]
webrtdsp: Remove restriction on channels number

Unlike 0.1, in 0.2 the reverse stream can have different number of
channels. Remove the check that restrict it.

8 years agowebrtcdsp: Style fix
Nicolas Dufresne [Thu, 23 Jun 2016 01:44:35 +0000 (21:44 -0400)]
webrtcdsp: Style fix

8 years agovtdec: minimize renegotiation to make hw decoding more reliable
Alessandro Decina [Thu, 16 Jun 2016 04:18:18 +0000 (14:18 +1000)]
vtdec: minimize renegotiation to make hw decoding more reliable

The hardware decoder can become (temporarily) unavailable across
VTDecompressionSessionCreate/Destroy calls. During negotiation if the currently
configured caps are still accepted by downstream we keep using them so we don't
have to destroy and recreate the decoding session.

This indirectly fixes https://bugzilla.gnome.org/show_bug.cgi?id=767429, by
making vtdec stick to GLMemory.

8 years agoqmlglsink: add win32 support
Matthew Waters [Thu, 16 Jun 2016 05:13:02 +0000 (15:13 +1000)]
qmlglsink: add win32 support

The current state of c++ ABI's on Window's and Gst's/Qt's conflicting
mingw builds means that we cannot use mingw for building the qt plugin.

Instead, a qmake .pro file is provided that is expected to be used with the
msvc binaries provided by Qt like so:

(with the PATH environment variable containing the path to the qt biniaries
and PKG_CONFIG_PATH containing the path to GStreamer modules)
cd /path/to/sources/gst-plugins-bad/ext/qt
qmake -tp vc

Then open the resulting VS project and build the library.  Then

cp debug/libgstqtsink.dll /path/to/prefix/lib/gstreamer-1.0/libgstqtsink.cll

https://bugzilla.gnome.org/show_bug.cgi?id=761260

8 years agoAdd missing file to gitignore
Nicolas Dufresne [Tue, 21 Jun 2016 15:40:22 +0000 (11:40 -0400)]
Add missing file to gitignore

8 years agodoc: Automatic update
Nicolas Dufresne [Mon, 20 Jun 2016 22:52:00 +0000 (18:52 -0400)]
doc: Automatic update

This is an automatic update with manual merges of running
"make update" in the doc/plugins directory. This should help
later maintenance of the plugins doc. A lot of plugin are
not referenced yet in the doc. Will come later.

8 years agodoc: Add .cpp files to doc parser
Nicolas Dufresne [Mon, 20 Jun 2016 20:30:19 +0000 (16:30 -0400)]
doc: Add .cpp files to doc parser

Also, complete the list of directory that need to be visited.

8 years agowebrtcdsp: Add WebRTC Audio Processing support
Nicolas Dufresne [Mon, 6 Jun 2016 19:10:05 +0000 (15:10 -0400)]
webrtcdsp: Add WebRTC Audio Processing support

This DSP library can be used to enhance voice signal for real time
communication call. In implements multiple filters like noise reduction,
high pass filter, echo cancellation, automatic gain control, etc.

The webrtcdsp element can be used along, or with the help of the
webrtcechoprobe if echo cancellation is enabled. The echo probe should
be placed as close as possible to the audio sink, while the DSP is
generally place close to the audio capture. For local testing, one can
use an echo loop pipeline like the following:

  autoaudiosrc ! webrtcdsp ! webrtcechoprobe ! autoaudiosink

This pipeline should produce a single echo rather then repeated echo.
Those elements works if they are placed in the same top level pipeline.

https://bugzilla.gnome.org/show_bug.cgi?id=767800

8 years agoAutomatic update of common submodule
Nicolas Dufresne [Tue, 21 Jun 2016 15:46:44 +0000 (11:46 -0400)]
Automatic update of common submodule

From ac2f647 to f363b32

8 years agoglcontext/wgl: try wglCreateContextAttribsARB even not gl3 core profile
Wang Xin-yu (王昕宇) [Tue, 22 Mar 2016 08:51:53 +0000 (16:51 +0800)]
glcontext/wgl: try wglCreateContextAttribsARB even not gl3 core profile

https://bugzilla.gnome.org/show_bug.cgi?id=764018

8 years agoglcontext: fix get_current_gl_api() on win32
Matthew Waters [Tue, 21 Jun 2016 08:28:46 +0000 (18:28 +1000)]
glcontext: fix get_current_gl_api() on win32

Another case of incorrect calling conventions.

Using this function on win32 would corrupt the stack pointer and end in
massive hilarity.

8 years agoglsl: fixup external-oes shaders by mangling the required extension
Matthew Waters [Wed, 15 Jun 2016 02:47:05 +0000 (12:47 +1000)]
glsl: fixup external-oes shaders by mangling the required extension

Newer devices require using a different GLSL extension for accessing
external-oes textures in a shader using the texture() functions.

While the GL_OES_EGL_image_external_essl3 should supposedly be supported
on a any GLES3 android device, the extension was defined after a lot of the
older drivers were built so they will not know about it.  Thus there are two
possible interpretations of which of texture[2D]() should be supported for
external-oes textures.  Strict adherence to the GL_OES_EGL_image_external
extension spec which uses texture2D() or following GLES3's pattern, also
allowing texture() as a function for accessing external-oes textures

This adds another mangling pass to convert
 #extension GL_OES_EGL_image_external : ...
into
 #extension GL_OES_EGL_image_external_essl3 : ...
on GLES3 and when the GL_OES_EGL_image_external_essl3 extension is supported.

Only uses texture() when the GLES3 and the GL_OES_EGL_image_external_essl3
extension is supported for external-oes textures.
Uses GLES2 + texture2D() + GL_OES_EGL_image_external in all other external-oes
cases.

https://bugzilla.gnome.org/show_bug.cgi?id=766993

8 years agowinks: Port to MSVC and pretend to be Windows XP
Nirbheek Chauhan [Tue, 24 May 2016 10:56:24 +0000 (16:26 +0530)]
winks: Port to MSVC and pretend to be Windows XP

strcasecmp is not defined on MSVC, so just use the glib wrapper. Also pretend to
be Windows XP explicitly since the API we use was deprecated and removed
(ifdef-ed) from the SDK after this version of Windows. This will be especially
relevant once we stop supporting Windows XP soon:

https://bugzilla.gnome.org/show_bug.cgi?id=756866

8 years agowin32: Add missing def files for libraries
Nirbheek Chauhan [Tue, 1 Mar 2016 20:47:54 +0000 (02:17 +0530)]
win32: Add missing def files for libraries

8 years agobuild: Factor out endian-order RGB formats
Nirbheek Chauhan [Tue, 1 Mar 2016 19:25:15 +0000 (00:55 +0530)]
build: Factor out endian-order RGB formats

MSVC seems to ignore preprocessor conditionals inside static pad templates

8 years agoopenjpegdec: use sampling field to determine RGB channel
Aaron Boxer [Tue, 14 Jun 2016 02:29:39 +0000 (22:29 -0400)]
openjpegdec: use sampling field to determine RGB channel

https://bugzilla.gnome.org/show_bug.cgi?id=767402

8 years agotemplatematch: fix buffer leak in test
Guillaume Desmottes [Thu, 2 Jun 2016 11:53:38 +0000 (13:53 +0200)]
templatematch: fix buffer leak in test

Need to call gst_check_drop_buffers() to unref the buffers exchanged
during the test.

https://bugzilla.gnome.org/show_bug.cgi?id=767159

8 years agoglimagesink: fix query leak in test
Guillaume Desmottes [Thu, 19 May 2016 10:31:56 +0000 (13:31 +0300)]
glimagesink: fix query leak in test

https://bugzilla.gnome.org/show_bug.cgi?id=766663

8 years agoglimagesink: properly setup/teardown tests
Guillaume Desmottes [Thu, 19 May 2016 10:29:15 +0000 (13:29 +0300)]
glimagesink: properly setup/teardown tests

tcase_add_checked_fixture() is suppose to call the setup and teardown
functions so the tests don't have to do it manually.

https://bugzilla.gnome.org/show_bug.cgi?id=766663

8 years agoglupload: fix shader leak in test
Guillaume Desmottes [Thu, 19 May 2016 10:09:41 +0000 (13:09 +0300)]
glupload: fix shader leak in test

https://bugzilla.gnome.org/show_bug.cgi?id=766663

8 years agouvch264demux: fix buffer leaks in test
Guillaume Desmottes [Thu, 19 May 2016 09:51:17 +0000 (12:51 +0300)]
uvch264demux: fix buffer leaks in test

The mock chain functions aren't unreffing the buffers so the caller
should do it.

https://bugzilla.gnome.org/show_bug.cgi?id=766663

8 years agocamerabin: fix element leak in test
Guillaume Desmottes [Thu, 19 May 2016 09:36:42 +0000 (12:36 +0300)]
camerabin: fix element leak in test

https://bugzilla.gnome.org/show_bug.cgi?id=766663

8 years agofix buffer leaks in tests
Guillaume Desmottes [Thu, 19 May 2016 08:19:57 +0000 (11:19 +0300)]
fix buffer leaks in tests

gst_check_drop_buffers() release the buffers exchanged during the test.

https://bugzilla.gnome.org/show_bug.cgi?id=766663

8 years agofix event leaks in tests
Guillaume Desmottes [Thu, 19 May 2016 08:19:20 +0000 (11:19 +0300)]
fix event leaks in tests

The events are supposed to be unreffed when finishing the test, not
reffed.

https://bugzilla.gnome.org/show_bug.cgi?id=766663

8 years agoaudiointerleave: fix message leaks by flushing the bus
Guillaume Desmottes [Thu, 19 May 2016 08:19:01 +0000 (11:19 +0300)]
audiointerleave: fix message leaks by flushing the bus

https://bugzilla.gnome.org/show_bug.cgi?id=766663

8 years agoaudioaggregator: fix buffer leak
Guillaume Desmottes [Thu, 19 May 2016 08:16:37 +0000 (11:16 +0300)]
audioaggregator: fix buffer leak

If the pad was still owning a buffer when being destroyed it was leaked.

Fix a leak with the test_flush_start_flush_stop test.

https://bugzilla.gnome.org/show_bug.cgi?id=766663