platform/upstream/gstreamer.git
9 years agoaudiovisualizers: merge audiovisualizer base classes
Luis de Bethencourt [Thu, 1 Oct 2015 14:35:37 +0000 (15:35 +0100)]
audiovisualizers: merge audiovisualizer base classes

These plugins now use the audiovisualizer base class in pbutils

https://bugzilla.gnome.org/show_bug.cgi?id=742875

9 years agogl: move debugging related functions to their own file
Matthew Waters [Thu, 1 Oct 2015 15:01:42 +0000 (01:01 +1000)]
gl: move debugging related functions to their own file

9 years agoglcolorscale: use glbasefilter vfuncs
Matthew Waters [Thu, 1 Oct 2015 13:20:19 +0000 (23:20 +1000)]
glcolorscale: use glbasefilter vfuncs

Uses less code \o/
Fixes legacy opengl rendering \o/

9 years agomssdemux: activate streams before configuring bitrate
Philippe Normand [Wed, 16 Sep 2015 10:50:46 +0000 (12:50 +0200)]
mssdemux: activate streams before configuring bitrate

Doing the contrary has no effect and the consequence is that playback
will start with the lowest bitrate even if we can already handle
higher bitrate.

https://bugzilla.gnome.org/show_bug.cgi?id=755108

9 years agomssdemux: PlayReady WRM parsing support
Philippe Normand [Mon, 8 Jun 2015 13:33:22 +0000 (15:33 +0200)]
mssdemux: PlayReady WRM parsing support

If the manifest has a ProtectionHeader node then parse it and emit
protection events according to the specified protection SystemID.

https://bugzilla.gnome.org/show_bug.cgi?id=753613

9 years agomssdemux: always use timescale for gst timestamp calculation
Philippe Normand [Thu, 3 Sep 2015 13:11:00 +0000 (15:11 +0200)]
mssdemux: always use timescale for gst timestamp calculation

Not doing this can lead the demuxer to attempt downloading fragments
for an invalid start time. The server would then send a HTTP
Precondition failed error, the demuxer would try some more times to
download the invalid fragment and eventually error out.

https://bugzilla.gnome.org/show_bug.cgi?id=754523

9 years agogl: Pass GL_CFLAGS to g-ir-scanner
Antoine Jacoutot [Wed, 30 Sep 2015 06:36:15 +0000 (08:36 +0200)]
gl: Pass GL_CFLAGS to g-ir-scanner

This unbreaks building when some headers are under a non-standard path.
e.g. /usr/X11R6/include as on OpenBSD.

https://bugzilla.gnome.org/show_bug.cgi?id=755850

9 years agoapplemedia: dist missing header files
Aleix Conchillo Flaqué [Wed, 30 Sep 2015 07:44:16 +0000 (00:44 -0700)]
applemedia: dist missing header files

https://bugzilla.gnome.org/show_bug.cgi?id=755852

9 years agoaudiointerleave: typecast bit-mask to guint64 to fix segmentation fault
Vineeth TM [Wed, 30 Sep 2015 04:13:19 +0000 (13:13 +0900)]
audiointerleave: typecast bit-mask to guint64 to fix segmentation fault

While creating caps in audiointerleave tests, bitmask is being set as 0x9
This is resulting in segmentation fault. Fix the same by typecasting to guint64

https://bugzilla.gnome.org/show_bug.cgi?id=755840

9 years agoglcontext: fixup strstr lengths so we don't overrun
Matthew Waters [Wed, 30 Sep 2015 05:17:38 +0000 (15:17 +1000)]
glcontext: fixup strstr lengths so we don't overrun

9 years agoglupload: remove useless release_buffer
Matthew Waters [Wed, 30 Sep 2015 03:31:50 +0000 (13:31 +1000)]
glupload: remove useless release_buffer

It's a leftover from when we weren't outputting GstBuffer's and
returning raw texture id's.

9 years agoglutils: use gst_element_set_context for setting display/other_context
Matthew Waters [Tue, 29 Sep 2015 15:53:53 +0000 (01:53 +1000)]
glutils: use gst_element_set_context for setting display/other_context

1. So we get tracking inside GstElement properly when e.g. adding to a bin
2. Removes redundant code.  Now only one place where
   GstContext->GstGLDisplay/GstGLContext transformation occurs
3. Fixes a memory leak in the process
4. Make the retrieval of debug categories thread safe

9 years agogtk: add some GL debug statements to show up in GL traces
Matthew Waters [Tue, 29 Sep 2015 12:57:52 +0000 (22:57 +1000)]
gtk: add some GL debug statements to show up in GL traces

9 years agohls: m3u8playlist: more clean-ups
Tim-Philipp Müller [Sun, 30 Aug 2015 16:11:31 +0000 (17:11 +0100)]
hls: m3u8playlist: more clean-ups

Remove unused functions; move GstM3U8 struct into .c file,
and indent the header so it's nice to read.

9 years agohls: m3u8playlist: fix silly queue iteration code
Tim-Philipp Müller [Sun, 30 Aug 2015 16:01:13 +0000 (17:01 +0100)]
hls: m3u8playlist: fix silly queue iteration code

9 years agohls: m3u8playlist: more minor playlist_render() clean-ups
Tim-Philipp Müller [Sun, 30 Aug 2015 15:40:02 +0000 (16:40 +0100)]
hls: m3u8playlist: more minor playlist_render() clean-ups

Move the TAG defines directly into the code, not sure what
their purposes is, these are printf format strings so having
them directly as literals in the code where they're used
makes the code easier to follow.

9 years agohls: m3u8playlist: move entry rendering into render_playlist()
Tim-Philipp Müller [Sun, 30 Aug 2015 15:34:26 +0000 (16:34 +0100)]
hls: m3u8playlist: move entry rendering into render_playlist()

Makes the code easier to follow and avoids unnecessary temporary
strings, since we can just append to the playlist GString directly.

9 years agohls: m3u8playlist: simplify gst_m3u8_playlist_render()
Tim-Philipp Müller [Sun, 30 Aug 2015 15:24:41 +0000 (16:24 +0100)]
hls: m3u8playlist: simplify gst_m3u8_playlist_render()

Remove playlist_str GString variable from GstM3U8Playlist struct,
since it's only used temporarily in playlist_render(). Might just
as well keep it local then.

9 years agohls: remove unused macro and direct gio usage
Tim-Philipp Müller [Sun, 30 Aug 2015 14:55:51 +0000 (15:55 +0100)]
hls: remove unused macro and direct gio usage

9 years agompegtsmux: fix downstream key unit events handling with hlssink
Tim-Philipp Müller [Tue, 29 Sep 2015 23:36:14 +0000 (00:36 +0100)]
mpegtsmux: fix downstream key unit events handling with hlssink

The buffer timestamps in the collect function will already be
running time, don't try to convert them again to running time,
this would yield CLOCK_TIME_NONE now that the segment is shifted
to account for negative dts.

This fixes x264enc ! mpegtsmux ! hlssink, which was broken
because mpegtsmux would send a downstream key unit event with
running time NONE and then hlssink would immediately send
another one upstream and it would just be a flood of force
keyframe events in both directions after the first one. This
would then break hlssink because it uses multifilesink in
next-file=key-unit-event mode, and starting a new file after
every few kB does not work well for HLS.

9 years agoopencv: erode: remove non-ip transform func
Luis de Bethencourt [Thu, 20 Aug 2015 16:35:04 +0000 (17:35 +0100)]
opencv: erode: remove non-ip transform func

Transform is set to be done in place in gstcvdilateerode.c, so the in-place
transform function is always used and the other is redundant. Removing it.

https://bugzilla.gnome.org/show_bug.cgi?id=753885

9 years agoopencv: dilate: remove non-ip transform func
Luis de Bethencourt [Thu, 20 Aug 2015 16:23:40 +0000 (17:23 +0100)]
opencv: dilate: remove non-ip transform func

Transform is set to be done in place in gstcvdilateerode.c, so the in-place
transform function is always used and the other is redundant. Removing it.

https://bugzilla.gnome.org/show_bug.cgi?id=753885

9 years agoopencv: use gst_opencv_video_filter_set_in_place()
Luis de Bethencourt [Tue, 29 Sep 2015 19:00:02 +0000 (20:00 +0100)]
opencv: use gst_opencv_video_filter_set_in_place()

Using the gst_base_transform function directly won't work. Need to use
gst_opencv_video_filter_set_in_place().

9 years agohanddetect: check for haar cascade files path
Luis de Bethencourt [Tue, 29 Sep 2015 16:00:22 +0000 (17:00 +0100)]
handdetect: check for haar cascade files path

When running GStreamer from uninstalled sources, the location of the haar
cascade files will be local. Check if running in uninstalled and set the
file paths accordingly.

9 years agoadaptivedemux: remove unused variable
John Chang [Sat, 26 Sep 2015 09:02:09 +0000 (17:02 +0800)]
adaptivedemux: remove unused variable

https://bugzilla.gnome.org/show_bug.cgi?id=755656

9 years agovideoaggregator: fix compilation with older glib version
Tim-Philipp Müller [Tue, 29 Sep 2015 12:31:18 +0000 (13:31 +0100)]
videoaggregator: fix compilation with older glib version

Remove weird use of private gtype defines and fix compilation
with older glib versions such as 2.36.

https://bugzilla.gnome.org/show_bug.cgi?id=755754

9 years agoopencv: cvlaplace: add scale and shift properties
Luis de Bethencourt [Tue, 29 Sep 2015 12:52:47 +0000 (13:52 +0100)]
opencv: cvlaplace: add scale and shift properties

Add scale and shift properties so these can be used to tweak the values
used in cvConvertScale() instead of using only defaults.

9 years agoqtsink: explicitely fallthrough switch statement
Luis de Bethencourt [Fri, 28 Aug 2015 15:24:24 +0000 (16:24 +0100)]
qtsink: explicitely fallthrough switch statement

In case ret is False, fallthrough to default case.

CID #1320705

9 years agovc1parse: add break to switch case
Luis de Bethencourt [Fri, 28 Aug 2015 15:13:16 +0000 (16:13 +0100)]
vc1parse: add break to switch case

Even though all cases inside VC1_STREAM_FORMAT_ASF are goto or
g_assert_not_reached(), add a break at the end to appease Coverity.

CID #1320706

9 years agoassrender: fix leak of tag samples in the tag list
Arnaud Vrac [Mon, 28 Sep 2015 18:56:56 +0000 (20:56 +0200)]
assrender: fix leak of tag samples in the tag list

Move handling of a GstSample in a separate function, and unref the
sample after calling it. libass copies the font data so we don't need to
keep it around.

https://bugzilla.gnome.org/show_bug.cgi?id=755759

9 years agoassrender: fix gap event leak and invalid return value
Arnaud Vrac [Mon, 28 Sep 2015 18:30:17 +0000 (20:30 +0200)]
assrender: fix gap event leak and invalid return value

We don't want the gap event to be forwarded

https://bugzilla.gnome.org/show_bug.cgi?id=755759

9 years agonvenc: call the parent class on GstElement::set_context
Matthew Waters [Tue, 29 Sep 2015 06:17:22 +0000 (16:17 +1000)]
nvenc: call the parent class on GstElement::set_context

9 years agogl/eagl: use the default GL context debug category
Matthew Waters [Thu, 17 Sep 2015 06:59:16 +0000 (16:59 +1000)]
gl/eagl: use the default GL context debug category

9 years agoAdd Nvidia based hardware encoder element
Matthew Waters [Thu, 30 Jul 2015 06:42:38 +0000 (16:42 +1000)]
Add Nvidia based hardware encoder element

Currently only h264 is supported

https://bugzilla.gnome.org/show_bug.cgi?id=753405

9 years agogtk: fix assertion when the element has no peer
Matthew Waters [Mon, 28 Sep 2015 14:25:00 +0000 (00:25 +1000)]
gtk: fix assertion when the element has no peer

When proxying keyboard/navigation/mouse events, only unref a successfully
retreived peer pad.

https://bugzilla.gnome.org/show_bug.cgi?id=755738

9 years agoRevert "midi: add an ALSA MIDI sequencer source"
Stefan Sauer [Mon, 28 Sep 2015 13:30:30 +0000 (15:30 +0200)]
Revert "midi: add an ALSA MIDI sequencer source"

This reverts commit cd7f4e524718a2f75467852235e4e13d64d2ac82.

9 years agogl: set the context on the element on a context query
Matthew Waters [Mon, 28 Sep 2015 12:31:09 +0000 (22:31 +1000)]
gl: set the context on the element on a context query

Otherwise it's possible to lose the context information if the
context is only propagated through queries.

9 years agogl: chain up to the parent class for GstElement::set_context
Matthew Waters [Mon, 28 Sep 2015 12:20:29 +0000 (22:20 +1000)]
gl: chain up to the parent class for GstElement::set_context

https://bugzilla.gnome.org/show_bug.cgi?id=705579

9 years agomidi: add an ALSA MIDI sequencer source
Antonio Ospite [Fri, 7 Aug 2015 13:28:54 +0000 (15:28 +0200)]
midi: add an ALSA MIDI sequencer source

The alsamidisrc element allows to get input event from ALSA MIDI
sequencer devices, and possibly convert them to sound using some
downstream element like fluiddec.

https://bugzilla.gnome.org/show_bug.cgi?id=738687

9 years agoqml: remove overwritten value
Luis de Bethencourt [Fri, 28 Aug 2015 15:35:39 +0000 (16:35 +0100)]
qml: remove overwritten value

Value in tex is overwritten before being used. Removing it.

CID 1320715

https://bugzilla.gnome.org/show_bug.cgi?id=754253

9 years agoqt: add support for building/running on android
Matthew Waters [Wed, 2 Sep 2015 13:45:07 +0000 (23:45 +1000)]
qt: add support for building/running on android

Including:
- Necessary configure checks
- Necessary compile time platform checks
- Necessary runtime qt android platform detection
- Escaping GLsync definition with Qt's GLES2 implementation

https://bugzilla.gnome.org/show_bug.cgi?id=754466

9 years agoqt: don't use CPPFLAGS for tools that cannot use them
Matthew Waters [Wed, 2 Sep 2015 13:40:31 +0000 (23:40 +1000)]
qt: don't use CPPFLAGS for tools that cannot use them

For example moc will bail out when given arguments it does not
know about.  The moc specific MOC_CPPFLAGS can still be used
to pass flags to moc.

https://bugzilla.gnome.org/show_bug.cgi?id=754466

9 years agoqt: rename library to include gst prefix
Matthew Waters [Wed, 2 Sep 2015 13:39:54 +0000 (23:39 +1000)]
qt: rename library to include gst prefix

libqtsink -> libgstqtsink

https://bugzilla.gnome.org/show_bug.cgi?id=754466

9 years agogl: sprinkle some debug markers to ease debugging
Matthew Waters [Mon, 31 Aug 2015 09:21:54 +0000 (19:21 +1000)]
gl: sprinkle some debug markers to ease debugging

9 years agogl/utils: add a function to insert a debug marker
Matthew Waters [Mon, 31 Aug 2015 09:18:23 +0000 (19:18 +1000)]
gl/utils: add a function to insert a debug marker

These markers are visible in tools that record the GL function calls
such as apitrace, et al.

Makes it easier to match up GL draw commands with specific elements.

9 years agogl: add some debugging prototypes
Matthew Waters [Mon, 31 Aug 2015 09:17:21 +0000 (19:17 +1000)]
gl: add some debugging prototypes

9 years agoglwindow: remove unused draw_unlocked function
Matthew Waters [Sat, 26 Sep 2015 09:19:50 +0000 (19:19 +1000)]
glwindow: remove unused draw_unlocked function

The functionality is provided by draw anyway and is leftover from
X11's specific threading requirements that no longer apply.

9 years agoresindvd: Send gap updates to the segment stop during stills
Jan Schmidt [Sun, 27 Sep 2015 03:07:19 +0000 (13:07 +1000)]
resindvd: Send gap updates to the segment stop during stills

Ignore the normal gap threshold for laggy streams and
immediately catch all streams up to the end of the segment
when processing gap updates for a segment during a
still frame sequence.

https://bugzilla.gnome.org/show_bug.cgi?id=755680

9 years agoRevert "dvdspu: render to AYUV overlay"
Jan Schmidt [Sat, 26 Sep 2015 14:24:24 +0000 (00:24 +1000)]
Revert "dvdspu: render to AYUV overlay"

This reverts commit 5016a73190595505dc38b54f8f9a08c4f180f3a6.

9 years agoRevert "dvdspu: render to ARGB overlay instead of AYUV"
Jan Schmidt [Sat, 26 Sep 2015 14:24:23 +0000 (00:24 +1000)]
Revert "dvdspu: render to ARGB overlay instead of AYUV"

This reverts commit dd3e9deb2aa695a391b58f24d86a3c00bbc1258a.

9 years agoRevert "dvdspu: handle frame size event from upstream"
Jan Schmidt [Sat, 26 Sep 2015 14:24:17 +0000 (00:24 +1000)]
Revert "dvdspu: handle frame size event from upstream"

This reverts commit 46aaaa6c309b8a8f6e218142fd5944f5d24f095c.

9 years agoRevert "dvdspu: cache overlay composition"
Jan Schmidt [Sat, 26 Sep 2015 14:24:17 +0000 (00:24 +1000)]
Revert "dvdspu: cache overlay composition"

This reverts commit aabb8a1a68372f750c63b9b6586e18e829745774.

9 years agoRevert "dvdspu: improve negotiation of overlay composition"
Jan Schmidt [Sat, 26 Sep 2015 14:24:15 +0000 (00:24 +1000)]
Revert "dvdspu: improve negotiation of overlay composition"

This reverts commit 1899e2a099932a7a734e7784c20b624517237e18.

9 years agodvdspu: fix pgs palette colors
Arnaud Vrac [Fri, 18 Sep 2015 17:07:47 +0000 (19:07 +0200)]
dvdspu: fix pgs palette colors

U and V were inverted

https://bugzilla.gnome.org/show_bug.cgi?id=755239

9 years agodvdspu: improve negotiation of overlay composition
Arnaud Vrac [Wed, 27 Aug 2014 14:47:22 +0000 (16:47 +0200)]
dvdspu: improve negotiation of overlay composition

Support negotiating GstVideoOverlayComposition downstream
while not providing it upstream.

https://bugzilla.gnome.org/show_bug.cgi?id=663750

9 years agodvdspu: cache overlay composition
Arnaud Vrac [Mon, 2 Sep 2013 15:48:50 +0000 (17:48 +0200)]
dvdspu: cache overlay composition

This avoids rendering the overlay buffer for each video frame.

https://bugzilla.gnome.org/show_bug.cgi?id=663750

9 years agodvdspu: handle frame size event from upstream
Arnaud Vrac [Fri, 30 Aug 2013 20:08:01 +0000 (22:08 +0200)]
dvdspu: handle frame size event from upstream

An IDX file or codec_data normally contains the original frame size of
the video. Allow upstream to provide this information by sending a
custom event, which will allow scaling the overlay correctly.

https://bugzilla.gnome.org/show_bug.cgi?id=663750

9 years agodvdspu: allow suffix in dvd event name to allow multiple sticky dvd events
Arnaud Vrac [Thu, 5 Dec 2013 20:59:18 +0000 (21:59 +0100)]
dvdspu: allow suffix in dvd event name to allow multiple sticky dvd events

https://bugzilla.gnome.org/show_bug.cgi?id=663750

9 years agodvdspu: render to ARGB overlay instead of AYUV
Arnaud Vrac [Tue, 7 Apr 2015 12:38:08 +0000 (14:38 +0200)]
dvdspu: render to ARGB overlay instead of AYUV

https://bugzilla.gnome.org/show_bug.cgi?id=663750

9 years agodvdspu: render to AYUV overlay
Arnaud Vrac [Wed, 23 Jan 2013 16:59:01 +0000 (17:59 +0100)]
dvdspu: render to AYUV overlay

Instead of only supporting writing SPU data directly to YUV frames,
render the SPU data to an intermediate AYUV overlay buffer. The overlay
data is then attached to the video frame if downstream supports overlay
composition, otherwise the AYUV overlay is blended to the video frame.

For the PGS format, the overlay buffer size is set to the size of the
Composition Window, and its position in the overlay composition is set
to the window position. The objects to render are now cropped when the
cropping flag is set.

For the Vobsub format, the overlay buffer size is set to the size of the
Display Area.

Once rendered, the overlay composition rectangle is now moved and scaled
to fit the video output size, to avoid clipping.

https://bugzilla.gnome.org/show_bug.cgi?id=663750

9 years agodvdspu: pass dvdspu argument to set_caps functions
Arnaud Vrac [Mon, 2 Sep 2013 15:07:49 +0000 (17:07 +0200)]
dvdspu: pass dvdspu argument to set_caps functions

https://bugzilla.gnome.org/show_bug.cgi?id=663750

9 years agodvdspu: skip unneeded reading of RLE data
Arnaud Vrac [Fri, 18 Sep 2015 17:15:45 +0000 (19:15 +0200)]
dvdspu: skip unneeded reading of RLE data

The RLE data was being read for dumping the SPU image even when the
DUMP_FULL_IMAGE macro was not set.

https://bugzilla.gnome.org/show_bug.cgi?id=663750

9 years agocheck: Add test for videoaggregator sinkpads being sorted by zorder
Nirbheek Chauhan [Sat, 29 Aug 2015 13:46:38 +0000 (19:16 +0530)]
check: Add test for videoaggregator sinkpads being sorted by zorder

https://bugzilla.gnome.org/show_bug.cgi?id=754285

9 years agoaudiomixer: fix deadlock when G_DISABLE_ASSERT is not defined
Tim-Philipp Müller [Sat, 26 Sep 2015 09:21:41 +0000 (10:21 +0100)]
audiomixer: fix deadlock when G_DISABLE_ASSERT is not defined

This makes the audiomixer unit test time out in master.
Broke with 587e7c4

9 years agodash: Fix unit test after internal API change
Sebastian Dröge [Sat, 26 Sep 2015 08:49:31 +0000 (10:49 +0200)]
dash: Fix unit test after internal API change

9 years agosfdec: Fix typo in goto variable name
Jimmy Ohn [Tue, 25 Aug 2015 08:39:30 +0000 (17:39 +0900)]
sfdec: Fix typo in goto variable name

https://bugzilla.gnome.org/show_bug.cgi?id=754057

9 years agodashdemux: create src pads for subtitle streams.
Chris Bass [Tue, 18 Aug 2015 13:16:11 +0000 (14:16 +0100)]
dashdemux: create src pads for subtitle streams.

Create src pads for Representations that contain timed-text subtitles,
both when the subtitles are encapsulated in ISO BMFF (i.e., the
Representation has mimeType "application/mp4") and when they are
unencapsulated (i.e., the Representation has mimeType
"application/ttml+xml").

https://bugzilla.gnome.org/show_bug.cgi?id=747774

9 years agoaudioaggregator: Stop using deprecated gst_segment_to_position()
Sebastian Dröge [Fri, 25 Sep 2015 22:17:55 +0000 (00:17 +0200)]
audioaggregator: Stop using deprecated gst_segment_to_position()

9 years agoaggregator: Don't forward QOS events to sinkpads that had no buffer yet
Sebastian Dröge [Mon, 31 Aug 2015 13:12:40 +0000 (16:12 +0300)]
aggregator: Don't forward QOS events to sinkpads that had no buffer yet

Otherwise they will receive a QOS event that has earliest_time=0 (because we
can't have negative timestamps), and consider their buffer as too late

https://bugzilla.gnome.org/show_bug.cgi?id=754356

9 years agodashdemux: Implement lazy-loading of external periods
Sebastian Dröge [Tue, 22 Sep 2015 14:17:38 +0000 (16:17 +0200)]
dashdemux: Implement lazy-loading of external periods

https://bugzilla.gnome.org/show_bug.cgi?id=752230

9 years agompdparser: Load OnLoad external resources immediately instead of on demand
Sebastian Dröge [Mon, 21 Sep 2015 19:05:03 +0000 (21:05 +0200)]
mpdparser: Load OnLoad external resources immediately instead of on demand

https://bugzilla.gnome.org/show_bug.cgi?id=752230

9 years agompdparser: Add support for loading external SegmentLists
Sebastian Dröge [Tue, 8 Sep 2015 10:36:23 +0000 (13:36 +0300)]
mpdparser: Add support for loading external SegmentLists

https://bugzilla.gnome.org/show_bug.cgi?id=752230

9 years agompdparser: Add support for loading external AdaptationSets
Sebastian Dröge [Tue, 8 Sep 2015 10:04:11 +0000 (13:04 +0300)]
mpdparser: Add support for loading external AdaptationSets

https://bugzilla.gnome.org/show_bug.cgi?id=752230

9 years agompdparser: Implement loading of external Period nodes
Sebastian Dröge [Fri, 21 Aug 2015 13:40:10 +0000 (16:40 +0300)]
mpdparser: Implement loading of external Period nodes

The same has to be done for AdaptationSet and SegmentList nodes still.

Also this does not correctly implement the semantics: by default Period (and
other nodes) should only be loaded when needed, not in the very beginning. We
need to implement lazy loading for them, which means adjusting
gst_mpd_client_setup_media_presentation().

https://bugzilla.gnome.org/show_bug.cgi?id=752230

9 years agompdparser: Store an URI downloader in the parser for downloading additional MPD resou...
Sebastian Dröge [Fri, 21 Aug 2015 09:06:07 +0000 (12:06 +0300)]
mpdparser: Store an URI downloader in the parser for downloading additional MPD resources if needed

https://bugzilla.gnome.org/show_bug.cgi?id=752230

9 years agompdparser: Parse xlink attributes from Period, AdaptationSet and SegmentList
Sebastian Dröge [Fri, 10 Jul 2015 15:56:29 +0000 (18:56 +0300)]
mpdparser: Parse xlink attributes from Period, AdaptationSet and SegmentList

We still have to do something useful with them, like actually loading the
content behind the URL.

https://bugzilla.gnome.org/show_bug.cgi?id=752230

9 years agodashdemux: Properly handle relative and non-HTTP URIs for the headers/indices
Sebastian Dröge [Wed, 16 Sep 2015 21:38:24 +0000 (23:38 +0200)]
dashdemux: Properly handle relative and non-HTTP URIs for the headers/indices

gst_uri_join_strings() will return the second parameter if it is an absolute
URI. No need to do a (wrong) check if the URI is absolute or not beforehand.

https://bugzilla.gnome.org/show_bug.cgi?id=755134

9 years agompdparser: If no Initialization is present in the SegmentBase, assume it is before...
Sebastian Dröge [Wed, 16 Sep 2015 21:12:54 +0000 (23:12 +0200)]
mpdparser: If no Initialization is present in the SegmentBase, assume it is before the indexRange

https://bugzilla.gnome.org/show_bug.cgi?id=755132

9 years agoBack to development
Sebastian Dröge [Fri, 25 Sep 2015 21:51:20 +0000 (23:51 +0200)]
Back to development

9 years agoRelease 1.6.0
Sebastian Dröge [Fri, 25 Sep 2015 21:33:37 +0000 (23:33 +0200)]
Release 1.6.0

9 years agoUpdate .po files
Sebastian Dröge [Fri, 25 Sep 2015 21:01:35 +0000 (23:01 +0200)]
Update .po files

9 years agoglfilter: use GL_ELEMENT_ARRAY_BUFFER for vbo indices
Julien Isorce [Fri, 25 Sep 2015 09:32:00 +0000 (10:32 +0100)]
glfilter: use GL_ELEMENT_ARRAY_BUFFER for vbo indices

Fixes this error with chromium gpu process:
GL_INVALID_OPERATION, glBindBuffer: buffer bound to more than 1 target

https://bugzilla.gnome.org/show_bug.cgi?id=755618

9 years agoglmemory: fix texture leak in _gl_mem_copy
Julien Isorce [Thu, 17 Sep 2015 13:17:09 +0000 (14:17 +0100)]
glmemory: fix texture leak in _gl_mem_copy

https://bugzilla.gnome.org/show_bug.cgi?id=755456

9 years agogtk: Only run from the main thread in stop() if we created the window
Sebastian Dröge [Thu, 24 Sep 2015 16:51:39 +0000 (18:51 +0200)]
gtk: Only run from the main thread in stop() if we created the window

We're not doing anything at all from the main thread in other cases.

9 years agogtk: When setting format check if pending format changed
Thibault Saunier [Thu, 24 Sep 2015 13:52:40 +0000 (15:52 +0200)]
gtk: When setting format check if pending format changed

In case the format changed fast and the pending format is different
than the currently set but the currently set is equal to the pending
one we could end up having mismatch between the finally set format
and the data stream format.

https://bugzilla.gnome.org/show_bug.cgi?id=755542

9 years agogtk: Do not forget to release OBJECT_LOCK on error path
Thibault Saunier [Thu, 24 Sep 2015 13:51:28 +0000 (15:51 +0200)]
gtk: Do not forget to release OBJECT_LOCK on error path

https://bugzilla.gnome.org/show_bug.cgi?id=755542

9 years agogtk: Factor out a function to run a function on main thread
Thibault Saunier [Thu, 24 Sep 2015 09:37:04 +0000 (11:37 +0200)]
gtk: Factor out a function to run a function on main thread

https://bugzilla.gnome.org/show_bug.cgi?id=755251

9 years agogtk: Marshall state changes in the main thread
Thibault Saunier [Thu, 24 Sep 2015 08:51:31 +0000 (10:51 +0200)]
gtk: Marshall state changes in the main thread

Gtk is not MT safe thus we need to make sure that everything is done
in the main thread when working with it.

https://bugzilla.gnome.org/show_bug.cgi?id=755251

9 years agodecklink: Add a clock epoch that is used as offset whenever restarting the clock
Sebastian Dröge [Wed, 23 Sep 2015 13:56:26 +0000 (15:56 +0200)]
decklink: Add a clock epoch that is used as offset whenever restarting the clock

Otherwise we're going to return times starting at 0 again after shutting down
an element for a specific input/output and then using it again later.

https://bugzilla.gnome.org/show_bug.cgi?id=755426

9 years agodecklink: Disable inputs/outputs in PAUSED->READY to allow going to PAUSED again...
Sebastian Dröge [Tue, 22 Sep 2015 17:35:00 +0000 (19:35 +0200)]
decklink: Disable inputs/outputs in PAUSED->READY to allow going to PAUSED again from there

https://bugzilla.gnome.org/show_bug.cgi?id=755426

9 years agogtksink: Do not show window until we reach the PAUSED state
Thibault Saunier [Wed, 23 Sep 2015 09:42:51 +0000 (11:42 +0200)]
gtksink: Do not show window until we reach the PAUSED state

https://bugzilla.gnome.org/show_bug.cgi?id=755459

9 years agompdparser: Only check stream->segments for a repeated last segment if we have a stati...
Sebastian Dröge [Tue, 22 Sep 2015 12:43:44 +0000 (14:43 +0200)]
mpdparser: Only check stream->segments for a repeated last segment if we have a static list of segments

Otherwise we'll crash, trying to derefence NULL. And if we have no static list
of segments, we can't have repeated segments anyway.

Regression introduced by cfe2871a5e3c5b1db20470927642b776e055a87a

https://bugzilla.gnome.org/show_bug.cgi?id=755411

9 years agox265enc: Drop incorrect stream-format="avc" in template caps
Arun Raghavan [Mon, 21 Sep 2015 13:32:08 +0000 (19:02 +0530)]
x265enc: Drop incorrect stream-format="avc" in template caps

9 years agogtksink: Do not re destroy the GtkWindow if destroyed by the user
Thibault Saunier [Mon, 21 Sep 2015 08:47:15 +0000 (10:47 +0200)]
gtksink: Do not re destroy the GtkWindow if destroyed by the user

Otherwise we will get an ASSERT.

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=755249

9 years agogl/dispmanx: fix rendering with recent resize state tracking
Matthew Waters [Mon, 21 Sep 2015 08:04:40 +0000 (08:04 +0000)]
gl/dispmanx: fix rendering with recent resize state tracking

557ca6fda5f831be4aba5819bf7b30b296e575cd didn't change to the
necessary gst_gl_window_resize() call for the dispmanx backend.

9 years agogtkglsink: Hide and clean the GtkWindow we might create
Thibault Saunier [Sat, 19 Sep 2015 09:46:37 +0000 (11:46 +0200)]
gtkglsink: Hide and clean the GtkWindow we might create

When stopping the sink we should always hide the window.

https://bugzilla.gnome.org/show_bug.cgi?id=755249

9 years agoRelease 1.5.91
Sebastian Dröge [Fri, 18 Sep 2015 17:51:49 +0000 (19:51 +0200)]
Release 1.5.91

9 years agoUpdate .po files
Sebastian Dröge [Fri, 18 Sep 2015 17:31:06 +0000 (19:31 +0200)]
Update .po files

9 years agoaudioaggregator: Only skip the remaining part of a GAP buffer
Sebastian Dröge [Fri, 18 Sep 2015 16:00:05 +0000 (18:00 +0200)]
audioaggregator: Only skip the remaining part of a GAP buffer

We might've queued up a GAP buffer that is only partially inside the current
output buffer (i.e. we received it too late!). In that case we should only
skip the part of the GAP buffer that is inside the current output buffer, not
also the remaining part. Otherwise we forward this pad too far into the future
and break synchronization.

9 years agovtdec: disable the texture cache on OSX
Matthew Waters [Fri, 18 Sep 2015 11:14:46 +0000 (21:14 +1000)]
vtdec: disable the texture cache on OSX

This a workaround until the incompatibility with CVOpenGLTextureCache
and opengl3 contexts is fixed.

https://bugzilla.gnome.org/show_bug.cgi?id=754786