platform/upstream/gstreamer.git
9 years agoaggregator: Keep at least two buffers in the queue in live mode
Olivier Crête [Thu, 17 Sep 2015 23:42:34 +0000 (19:42 -0400)]
aggregator: Keep at least two buffers in the queue in live mode

When in live mode, the queue needs to hold the currently processed
buffer and one more at least.

https://bugzilla.gnome.org/show_bug.cgi?id=754851

9 years agopo: Update translations
Sebastian Dröge [Fri, 18 Sep 2015 09:55:37 +0000 (11:55 +0200)]
po: Update translations

9 years agovtdec: gst-indent file
Matthew Waters [Fri, 18 Sep 2015 05:28:27 +0000 (15:28 +1000)]
vtdec: gst-indent file

9 years agodvdspu: pgs: fix printf format/argument mismatch in debug statements
Tim-Philipp Müller [Thu, 17 Sep 2015 18:56:47 +0000 (19:56 +0100)]
dvdspu: pgs: fix printf format/argument mismatch in debug statements

Fixes compiler warnings when extra debugging is enabled at compile time.

9 years agoadaptivedemux: remove unprefix STATISTICS_MESSAGE_NAME after all
Tim-Philipp Müller [Thu, 17 Sep 2015 16:54:36 +0000 (17:54 +0100)]
adaptivedemux: remove unprefix STATISTICS_MESSAGE_NAME after all

It was only added during 1.5.x so we might just as well remove it
immediately.

https://bugzilla.gnome.org/show_bug.cgi?id=754686

9 years agoadaptivedemux: prefix the statistics message name macro
Philippe Normand [Mon, 7 Sep 2015 14:57:05 +0000 (16:57 +0200)]
adaptivedemux: prefix the statistics message name macro

Keep old define around for now.

https://bugzilla.gnome.org/show_bug.cgi?id=754686

9 years agopcapparse: Hook up the recently added change_state vfunc.
Jan Schmidt [Thu, 17 Sep 2015 14:10:10 +0000 (00:10 +1000)]
pcapparse: Hook up the recently added change_state vfunc.

The state change function was implemented, but not installed,
making the compiler complain about the unused function.

9 years agoDon't throw compiler warnings with G_DISABLE_ASSERT
Jan Schmidt [Thu, 17 Sep 2015 14:20:13 +0000 (00:20 +1000)]
Don't throw compiler warnings with G_DISABLE_ASSERT

Disable code that warns about unused variables when G_DISABLE_ASSERT
is defined, as it is in tarballs and pre-releases.

9 years agogl/caopengllayer: fix non-existent selector warning
Matthew Waters [Thu, 17 Sep 2015 12:17:24 +0000 (22:17 +1000)]
gl/caopengllayer: fix non-existent selector warning

557ca6fda5f831be4aba5819bf7b30b296e575cd introduced the queueResize
call without implementing the selector

9 years agoglwindow: Fix g_return_val_if_fail in a void function
Jan Schmidt [Thu, 17 Sep 2015 11:31:39 +0000 (21:31 +1000)]
glwindow: Fix g_return_val_if_fail in a void function

9 years agoapplemedia/vtdec: fix negotiation errors when resizing glimagesink
Matthew Waters [Thu, 17 Sep 2015 05:06:31 +0000 (15:06 +1000)]
applemedia/vtdec: fix negotiation errors when resizing glimagesink

GstVideoDecoder has its own logic for detecting when to reconfigure
which ultimately calls decide_allocation and results in a new
texture cache that has not been configured from our reconfigure check.

https://bugzilla.gnome.org/show_bug.cgi?id=755156

9 years agoglimagesink: request a resize on caps/3d mode changes
Matthew Waters [Thu, 17 Sep 2015 07:22:47 +0000 (17:22 +1000)]
glimagesink: request a resize on caps/3d mode changes

Fixes incorrect aspect ratio on OSX when changing caps or the 3d
output mode.

https://bugzilla.gnome.org/show_bug.cgi?id=755111

9 years agoglwindow: add API to request a resize event on the next draw
Matthew Waters [Thu, 17 Sep 2015 07:06:37 +0000 (17:06 +1000)]
glwindow: add API to request a resize event on the next draw

- glimagesink needs to be able to resize the viewport on aspect ratio
  changes resulting from either caps changes or 3d output mode changes.
- Performing a glViewport outside the GstGLWindow::resize callback
  will not have the winsys' stack of viewports required to correctly
  place the output frame.

Provide a function to request a resize on the next draw event from the
winsys.

Also track size changes inside the base GstGLWindow class rather
than in each subclass.

https://bugzilla.gnome.org/show_bug.cgi?id=755111

9 years agogl/uploadelement: fail earlier if we could not upload the buffer
Matthew Waters [Thu, 17 Sep 2015 06:55:11 +0000 (16:55 +1000)]
gl/uploadelement: fail earlier if we could not upload the buffer

9 years agogl/x11: store the correct dimension from the resize events
Matthew Waters [Thu, 17 Sep 2015 06:05:21 +0000 (16:05 +1000)]
gl/x11: store the correct dimension from the resize events

small typo s/width/height/

9 years agoavdtputil: fix GError memory leak during failures
Vineeth T M [Thu, 17 Sep 2015 04:47:23 +0000 (13:47 +0900)]
avdtputil: fix GError memory leak during failures

https://bugzilla.gnome.org/show_bug.cgi?id=755142

9 years agogl: Fix GError leaks during failures
Vineeth T M [Thu, 17 Sep 2015 04:35:02 +0000 (13:35 +0900)]
gl: Fix GError leaks during failures

https://bugzilla.gnome.org/show_bug.cgi?id=755140

9 years agointervideosink: render preroll frames as well
Tim-Philipp Müller [Tue, 15 Sep 2015 23:19:35 +0000 (00:19 +0100)]
intervideosink: render preroll frames as well

Derive from GstVideoSink so that preroll frames will automatically
get rendered too, unless the show-preroll-frame property is set to
FALSE. Fixes intervideosrc only picking up frames if intervideosink
is in PLAYING state.

https://bugzilla.gnome.org/show_bug.cgi?id=755049

9 years agodvbsuboverlay: Check if downstream supports GstVideoOverlayComposition.
Jan Schmidt [Wed, 16 Sep 2015 14:01:58 +0000 (00:01 +1000)]
dvbsuboverlay: Check if downstream supports GstVideoOverlayComposition.

Fix the negotiation of GstVideoOverlayComposition by checking
intersection with the peer caps, rather than just accept-caps,
which might only check the pad template.

https://bugzilla.gnome.org/show_bug.cgi?id=755113

9 years agompdparser: support for negative repeat count in segments
Vincent Penquerc'h [Wed, 9 Sep 2015 13:49:17 +0000 (14:49 +0100)]
mpdparser: support for negative repeat count in segments

Implements negative repeat segment fields, defined in 5.3.9.6.1.

9 years agompdparser: properly read signed r values for S elements
Vincent Penquerc'h [Tue, 8 Sep 2015 14:14:13 +0000 (15:14 +0100)]
mpdparser: properly read signed r values for S elements

The spec defines these as signed in 5.3.9.6.1.
Since we don't support this behavior, warn and default to 0
(non repeating), which is the spec's default when the value
is not present.

https://bugzilla.gnome.org/show_bug.cgi?id=752480

9 years agomdpparser: catch failures to parse
Vincent Penquerc'h [Wed, 9 Sep 2015 10:05:35 +0000 (11:05 +0100)]
mdpparser: catch failures to parse

https://bugzilla.gnome.org/show_bug.cgi?id=752480

9 years agoglimagesink: avoid updating the viewport in the draw loop
Matthew Waters [Wed, 16 Sep 2015 07:28:05 +0000 (17:28 +1000)]
glimagesink: avoid updating the viewport in the draw loop

Fixes flashes/incorrect aspect ratio when resizing glimagesink on OSX.

9 years agoadaptivedemux: Fix playback of live streams
Sebastian Dröge [Tue, 15 Sep 2015 19:50:19 +0000 (21:50 +0200)]
adaptivedemux: Fix playback of live streams

dashdemux seeks each live stream to its current fragment in the beginning, but
the base class does not know about this. Update the demuxer segment with this
seek so we generate the correct SEGMENT event and can actually play the
stream.

This needs some refactoring at some point.

https://bugzilla.gnome.org/show_bug.cgi?id=755047

9 years agoopusdec: remove check for number of channels
Thiago Santos [Tue, 15 Sep 2015 18:39:11 +0000 (15:39 -0300)]
opusdec: remove check for number of channels

opus decoder can convert from different number of channels, no
need to check, just let it negotiate and create a new decoder if
needed.

https://bugzilla.gnome.org/show_bug.cgi?id=755059

9 years agoadaptivedemux: Only overwrite last return if it was OK
Thiago Santos [Tue, 15 Sep 2015 13:00:43 +0000 (10:00 -0300)]
adaptivedemux: Only overwrite last return if it was OK

Prevents overwriting other conditions that would be more important,
such as EOS.

https://bugzilla.gnome.org/show_bug.cgi?id=755042

9 years agodashdemux: fix return when trying to advance in subfragmented stream
Thiago Santos [Tue, 15 Sep 2015 12:58:20 +0000 (09:58 -0300)]
dashdemux: fix return when trying to advance in subfragmented stream

Even if it doesn't actually advance the subfragment in the default way
for streams that have subfragments, it can help the base class to return
EOS when there is no more fragments instead of signaling it that it should
continue downloading.

https://bugzilla.gnome.org/show_bug.cgi?id=755042

9 years agogl: bufferpool take into account video stride alignment requirements
Tim-Philipp Müller [Tue, 15 Sep 2015 10:34:12 +0000 (11:34 +0100)]
gl: bufferpool take into account video stride alignment requirements

when allocating memory. Fixes crashes with avdec_h265 in the AVX2
code path which requires 32-byte stride alignment, but the
GstAllocationParams only specified a 16-byte alignment.

https://bugzilla.gnome.org/show_bug.cgi?id=754120

9 years agopitch: Set seqnum on newly created segment event
Thibault Saunier [Tue, 15 Sep 2015 11:40:58 +0000 (13:40 +0200)]
pitch: Set seqnum on newly created segment event

https://bugzilla.gnome.org/show_bug.cgi?id=755012

9 years agoaudioaggregator: Fix mixup of running times and segment positions
Sebastian Dröge [Fri, 11 Sep 2015 19:37:08 +0000 (21:37 +0200)]
audioaggregator: Fix mixup of running times and segment positions

We have to queue buffers based on their running time, not based on
the segment position.

Also return running time from GstAggregator::get_next_time() instead of
a segment position, as required by the API.

Also only update the segment position after we pushed a buffer, otherwise
we're going to push down a segment event with the next position already.

https://bugzilla.gnome.org/show_bug.cgi?id=753196

9 years agoaudioaggregator: Use stream time in the position query instead of segment position
Sebastian Dröge [Fri, 11 Sep 2015 14:56:40 +0000 (16:56 +0200)]
audioaggregator: Use stream time in the position query instead of segment position

https://bugzilla.gnome.org/show_bug.cgi?id=753196

9 years agovideoaggregator: Fix mixup of running times and segment positions
Sebastian Dröge [Fri, 11 Sep 2015 10:22:51 +0000 (12:22 +0200)]
videoaggregator: Fix mixup of running times and segment positions

We have to queue buffers based on their running time, not based on
the segment position.

Also return running time from GstAggregator::get_next_time() instead of
a segment position, as required by the API.

Also only update the segment position after we pushed a buffer, otherwise
we're going to push down a segment event with the next position already.

https://bugzilla.gnome.org/show_bug.cgi?id=753196

9 years agoaggregator: Document that get_next_time() should return running time
Sebastian Dröge [Fri, 11 Sep 2015 10:21:50 +0000 (12:21 +0200)]
aggregator: Document that get_next_time() should return running time

https://bugzilla.gnome.org/show_bug.cgi?id=753196

9 years agompdparser: Don't consider period start times in periods with segment lists either
Sebastian Dröge [Thu, 3 Sep 2015 11:20:00 +0000 (14:20 +0300)]
mpdparser: Don't consider period start times in periods with segment lists either

https://bugzilla.gnome.org/show_bug.cgi?id=754222

9 years agompdparser: Fix unit test that assumed that fragment timestamps should include the...
Sebastian Dröge [Thu, 3 Sep 2015 07:26:03 +0000 (10:26 +0300)]
mpdparser: Fix unit test that assumed that fragment timestamps should include the period start timestamp

https://bugzilla.gnome.org/show_bug.cgi?id=754222

9 years agodashdemux: Export the period start time to the base class
Sebastian Dröge [Wed, 2 Sep 2015 15:33:51 +0000 (18:33 +0300)]
dashdemux: Export the period start time to the base class

https://bugzilla.gnome.org/show_bug.cgi?id=754222

9 years agoadaptivedemux: Properly implement timestamping of multi-period streams
Sebastian Dröge [Wed, 2 Sep 2015 15:29:43 +0000 (18:29 +0300)]
adaptivedemux: Properly implement timestamping of multi-period streams

Each period will start again with pts 0 + period presentation offset, which is
also going to be the presentation time inside the container stream if any.
However all periods together should form a continuous timeline, with regard to
stream time and running time.

For making this possible we keep track of the "user requested segment", i.e.
the seek events, inside the demuxer without adjusting anything and taking this
demuxer segment only as orientation for modified segments per stream.

This per stream segments will have their segment.start at pts that would be
produced for this stream in this period, and the segment.base/time adjusted so
that this pts maps to the running and stream time this period should have in
the context of all other periods.

https://bugzilla.gnome.org/show_bug.cgi?id=754222

9 years agoRevert "dashdemux: Subtract the period start time from the presentation offset"
Sebastian Dröge [Tue, 1 Sep 2015 10:13:58 +0000 (13:13 +0300)]
Revert "dashdemux: Subtract the period start time from the presentation offset"

This reverts commit 626a8f0a74f8ea748b811b74ba9e7ae2baea2cca.

This allows us to get the plain presentation offset and the period start time
separately. We have to adjust the timestamp by the presentation offset, but
the period start time should only adjust the stream time and running time.

https://bugzilla.gnome.org/show_bug.cgi?id=752409

9 years agoRevert "dashdemux: Include the period start in the fragment timestamps in all cases"
Sebastian Dröge [Tue, 1 Sep 2015 10:12:45 +0000 (13:12 +0300)]
Revert "dashdemux: Include the period start in the fragment timestamps in all cases"

This reverts commit e671ad25a989cb21c62c7a5867c2090890ce49ba.

The timestamps should restart at 0 again for each period, but we have to
adjust the segment to map those timestamps to the actual stream time and
running time of that period.

Otherwise we would have timestamps that conflict with the ones from the tfdt
inside the MP4 container, which are restarting at 0 for each period.

https://bugzilla.gnome.org/show_bug.cgi?id=752409

9 years agompdparser: Consider the repeat count when checking if a segment is the last one
Sebastian Dröge [Mon, 14 Sep 2015 17:03:10 +0000 (19:03 +0200)]
mpdparser: Consider the repeat count when checking if a segment is the last one

Otherwise we play only the first repetition of the last segment and then EOS.

https://bugzilla.gnome.org/show_bug.cgi?id=752480

9 years agodashdemux: also check for subfragments on has_next_fragment
Thiago Santos [Mon, 14 Sep 2015 16:56:10 +0000 (13:56 -0300)]
dashdemux: also check for subfragments on has_next_fragment

In dash isombff profile the fragment is split into subframents where
bitrate switching is possible. Also take that into consideration
when checking if a stream has next fragments.

9 years agoopusenc: improve deprecated properties docs
Miguel París Díaz [Fri, 11 Sep 2015 09:22:35 +0000 (11:22 +0200)]
opusenc: improve deprecated properties docs

https://bugzilla.gnome.org/show_bug.cgi?id=754819

9 years agoopusenc: do not throw g_warning when getting deprecated properties
Miguel París Díaz [Fri, 11 Sep 2015 09:11:09 +0000 (11:11 +0200)]
opusenc: do not throw g_warning when getting deprecated properties

https://bugzilla.gnome.org/show_bug.cgi?id=754819

9 years agovideoaggregator: fix caps query to properly handle alpha formats
Thiago Santos [Wed, 9 Sep 2015 22:51:18 +0000 (19:51 -0300)]
videoaggregator: fix caps query to properly handle alpha formats

Only accept alpha if downstream has alpha as well. It could
theoretically accept alpha unconditionally if blending is
properly implemented for handle it but at the moment this
is a missing feature.

Improves the caps query by also comparing with the template
caps to filter by what the subclass supports.

https://bugzilla.gnome.org/show_bug.cgi?id=754465

9 years agovtdec: Use R/RG textures for NV12 when the context provides them
Jan Schmidt [Fri, 11 Sep 2015 14:51:52 +0000 (00:51 +1000)]
vtdec: Use R/RG textures for NV12 when the context provides them

Fixes playback to GL memory on iOS, where the colours are messed
up by passing Luminance/LuminanceAlpha textures where
color convert expects R/RG textures.

https://bugzilla.gnome.org/show_bug.cgi?id=754504

9 years agovtdec: Prefer GL output caps.
Jan Schmidt [Tue, 8 Sep 2015 07:17:36 +0000 (17:17 +1000)]
vtdec: Prefer GL output caps.

Prefer GL caps by putting them first in the pad template and
intersecting using GST_CAPS_INTERSECT_FIRST.

9 years agocodecparsers: h265: Fix the selection of Active Ref Pic Set
Sreerenj Balachandran [Thu, 10 Sep 2015 16:02:12 +0000 (19:02 +0300)]
codecparsers: h265: Fix the selection of Active Ref Pic Set

If short_term_ref_pic_set_sps_flag is FALSE, the ShortTermRefPicSet
structure is supposed to derive from slice header. Which means,
CurrRpsIdx is equal to num_short_term_ref_pic_sets. But the number
of refpicsets communicated via sps header is only num_short_term_ref_pic_sets - 1.
And we are using slice_header structure to reference the last entry, which is
ShortTermRefPicSet[num_short_term_ref_pic_sets].

https://bugzilla.gnome.org/show_bug.cgi?id=754834

9 years agocodecparsers: h265 : Fix default scaling list values
Sreerenj Balachandran [Thu, 10 Sep 2015 13:28:31 +0000 (16:28 +0300)]
codecparsers: h265 : Fix default scaling list values

Fix the default_scaling_list values based on Table 7-6

https://bugzilla.gnome.org/show_bug.cgi?id=754834

9 years agodecklinkvideosink: less-than-zero comparison of an unsigned value is never true
Sebastian Dröge [Thu, 10 Sep 2015 11:11:58 +0000 (14:11 +0300)]
decklinkvideosink: less-than-zero comparison of an unsigned value is never true

Use the correct type, GstClockTimeDiff, instead.

CID 1323742

9 years agogtk, qt: more specifically define the compile time requirements
Matthew Waters [Wed, 9 Sep 2015 14:07:18 +0000 (00:07 +1000)]
gtk, qt: more specifically define the compile time requirements

Otherwise we could include headers/configurations that will
never been installed.

https://bugzilla.gnome.org/show_bug.cgi?id=754732

9 years agobuild/qt: indent configure file
Matthew Waters [Wed, 2 Sep 2015 13:42:34 +0000 (23:42 +1000)]
build/qt: indent configure file

don't mix tabs and spaces

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

9 years agoqt: use our function table instead of directly calling gl functions
Matthew Waters [Wed, 9 Sep 2015 14:00:11 +0000 (00:00 +1000)]
qt: use our function table instead of directly calling gl functions

Otherwise when building with --as-needed we would need to link to
a GL or GLES library.

https://bugzilla.gnome.org/show_bug.cgi?id=754732

9 years agogl/api: use public rather than private define
Matthew Waters [Wed, 9 Sep 2015 13:53:57 +0000 (23:53 +1000)]
gl/api: use public rather than private define

HAVE_IOS is only defined for the build of this module so
attempting to use gstgl in iOS would result in incorrect GL
includes.
Use GST_GL_HAVE_PLATFORM_EAGL instead for choosing the iOS GL
header.

9 years agodecklinkvideosink: Handle pipelines where the running time does not start around...
Sebastian Dröge [Thu, 3 Sep 2015 13:36:57 +0000 (16:36 +0300)]
decklinkvideosink: Handle pipelines where the running time does not start around 0 properly

We were converting all times to our internal running times, that is the time
the sink itself spent in PLAYING already. But forgot to do that for the
running time calculated from the buffer timestamps. As such, all buffers were
scheduled much later if the pipeline's running time did not start at 0.

This happens for example if a base time is explicitly set on the pipeline.

https://bugzilla.gnome.org/show_bug.cgi?id=754528

9 years agodecklinkvideosink: Consider pipeline latency, render delay and ts offset when schedul...
Sebastian Dröge [Mon, 7 Sep 2015 10:36:19 +0000 (13:36 +0300)]
decklinkvideosink: Consider pipeline latency, render delay and ts offset when scheduling frames

Without this, we will schedule all frames too late in live pipelines.

https://bugzilla.gnome.org/show_bug.cgi?id=754666

9 years agodirectsoundsrc: fix build error for 64bit Windows build by MinGW
Kouhei Sutou [Wed, 9 Sep 2015 04:58:05 +0000 (13:58 +0900)]
directsoundsrc: fix build error for 64bit Windows build by MinGW

Casting to UINT from HMIXER generates the following warning with
64bit Windows target MinGW:

    gstdirectsoundsrc.c: In function 'gst_directsound_src_mixer_find':
    gstdirectsoundsrc.c:733:30: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]
         mmres = mixerGetDevCaps ((UINT) dsoundsrc->mixer,
                                  ^
    cc1: all warnings being treated as errors

We can use portable GPOINTER_TO_UINT() macro for this propose.

https://bugzilla.gnome.org/show_bug.cgi?id=754756

9 years agogl/eagl: Also unbind renderbuffer after setting up framebuffer
Sebastian Dröge [Wed, 9 Sep 2015 06:33:01 +0000 (09:33 +0300)]
gl/eagl: Also unbind renderbuffer after setting up framebuffer

https://bugzilla.gnome.org/show_bug.cgi?id=754757

9 years agogl/eagl: Don't make context uncurrent just to make it current again in the next line
Sebastian Dröge [Wed, 9 Sep 2015 06:27:17 +0000 (09:27 +0300)]
gl/eagl: Don't make context uncurrent just to make it current again in the next line

Also binding the framebuffer again is unnecessary then as it was just bound a
few lines before while the context was current.

https://bugzilla.gnome.org/show_bug.cgi?id=754757

9 years agompdparser: fix period start time scaling
Vincent Penquerc'h [Mon, 7 Sep 2015 11:09:41 +0000 (12:09 +0100)]
mpdparser: fix period start time scaling

This GstStreamPeriod start value is expressed in nanoseconds,
and the glib time addition function expects microseconds.
There seems to have been a confusion with GstPeriodNode's start
field, which is expressed in milliseconds.

Additionally, add a warning if the timestamp modification did
not succeed, and NULL was returned.

9 years agofaac: Add bitrate info to the tags output.
Jan Schmidt [Sat, 5 Sep 2015 13:38:44 +0000 (23:38 +1000)]
faac: Add bitrate info to the tags output.

Makes it possible for muxers to know the target bitrate as soon
as encoding starts, which flvmux now uses.

9 years agocompositor: Ensure all arguments to CLAMP are signed int
Nirbheek Chauhan [Fri, 4 Sep 2015 20:20:41 +0000 (01:50 +0530)]
compositor: Ensure all arguments to CLAMP are signed int

If any of the arguments to CLAMP are unsigned integers, the comparison causes
an automatic conversion of the signed int to unsigned, which causes -1 to become
UINT_MAX and get clamped to the high value of the CLAMP instead of 0.

See 716 at http://c0x.coding-guidelines.com/6.3.1.8.html

Also add a test for this.

https://bugzilla.gnome.org/show_bug.cgi?id=754576

9 years agogtk: Do not consider GtkEvents as handled
Thibault Saunier [Fri, 4 Sep 2015 13:43:40 +0000 (15:43 +0200)]
gtk: Do not consider GtkEvents as handled

Applications might still want to use them
after the sink transformed them into
GstNavigation events

9 years agovideoaggregator: lift restriction of changing pixel-aspect-ratio
Thiago Santos [Wed, 2 Sep 2015 22:16:56 +0000 (19:16 -0300)]
videoaggregator: lift restriction of changing pixel-aspect-ratio

The videoaggregator can convert PAR, there is no reason for
restricting it.

https://bugzilla.gnome.org/show_bug.cgi?id=754291

9 years agoqmlsink: Ensure that at least one windowing system is available
Nirbheek Chauhan [Fri, 7 Aug 2015 11:57:48 +0000 (17:27 +0530)]
qmlsink: Ensure that at least one windowing system is available

Otherwise, we'll just crash at runtime because the gl context is NULL

https://bugzilla.gnome.org/show_bug.cgi?id=754108

9 years agopcapparse: Reset state on flush and state changes
Olivier Crête [Sat, 29 Aug 2015 00:33:40 +0000 (20:33 -0400)]
pcapparse: Reset state on flush and state changes

https://bugzilla.gnome.org/show_bug.cgi?id=753152

9 years agogtk, qt, gl: fix typo in debug and error messages
Tim-Philipp Müller [Mon, 31 Aug 2015 17:06:31 +0000 (18:06 +0100)]
gtk, qt, gl: fix typo in debug and error messages

9 years agoaggregator: Also ignore start-time on seek from gst_element_send_event()
Olivier Crête [Sat, 29 Aug 2015 03:05:20 +0000 (23:05 -0400)]
aggregator: Also ignore start-time on seek from gst_element_send_event()

https://bugzilla.gnome.org/show_bug.cgi?id=753806

9 years agocompositor: variables in clamp_rectangle() should be signed
Nirbheek Chauhan [Sat, 29 Aug 2015 11:21:08 +0000 (16:51 +0530)]
compositor: variables in clamp_rectangle() should be signed

x/y/w/h are signed integers. As can be seen in GstCompositorPad.
The prototype for clamp_rectangle was wrong. This commit reverts the change
and fixes the prototype.

This reverts commit bca444ea4a84c39e9989681f892f6e4cb2033cf9.

9 years agoFix file permissions of some files
Tim-Philipp Müller [Fri, 28 Aug 2015 18:55:59 +0000 (19:55 +0100)]
Fix file permissions of some files

9 years agocompositor: remove check for below zero for unsigned value
Luis de Bethencourt [Fri, 28 Aug 2015 14:21:11 +0000 (15:21 +0100)]
compositor: remove check for below zero for unsigned value

CLAMP checks both if value is '< 0' and '> max'. Value will never be a negative
number since it is an unsigned integer. Removing that check and only checking if
it is bigger than max by using MIN().

CID 1320707

9 years agoedgedetect: Rename gstedgedetect.c to gstedgedetect.cpp for consistency.
Vanessa Chipirrás Navalón [Fri, 28 Aug 2015 11:42:29 +0000 (13:42 +0200)]
edgedetect: Rename gstedgedetect.c to gstedgedetect.cpp for consistency.

Change the file extension to cpp and add it into Makefile for consistency
with other elements of opencv.

https://bugzilla.gnome.org/show_bug.cgi?id=754148

9 years agoedgedetect: remove unused and useless functions.
Vanessa Chipirrás Navalón [Thu, 27 Aug 2015 19:51:52 +0000 (21:51 +0200)]
edgedetect: remove unused and useless functions.

The cvSmooth cvNot functions and do not have the correct input parameters.
Furthermore, cvSmooth function is not necessary for edge detection,
because the Canny function makes the step of smoothing the image.
And cvNot function is useless because there aren't changes if this
function is eliminated.

https://bugzilla.gnome.org/show_bug.cgi?id=754148

9 years agotextoverlay: remove unused variable
Luis de Bethencourt [Wed, 26 Aug 2015 17:40:35 +0000 (18:40 +0100)]
textoverlay: remove unused variable

Memory is reserved for this variable but never used. Removing it.

9 years agofaceblur: remove unused variable.
Vanessa Chipirrás Navalón [Wed, 26 Aug 2015 16:21:15 +0000 (18:21 +0200)]
faceblur: remove unused variable.

Memory is reserved for this variable and then released without making any
use of it.

https://bugzilla.gnome.org/show_bug.cgi?id=753994

9 years agofaceblur: need to migrate to C++.
Vanessa Chipirrás Navalón [Wed, 26 Aug 2015 15:56:50 +0000 (17:56 +0200)]
faceblur: need to migrate to C++.

The cascade classifier changes its structure on new version of OpenCV 2.4.11.
It is need to migrate to C++ to utilize the new load method of OpenCV which
allows to load the old and new classifiers.

https://bugzilla.gnome.org/show_bug.cgi?id=753994

9 years agofaceblur: Check CvHaarClassifierCascade is release before being modified.
Vanessa Chipirrás Navalón [Wed, 26 Aug 2015 10:47:58 +0000 (12:47 +0200)]
faceblur: Check CvHaarClassifierCascade is release before being modified.

For PROP_PROFILE case that exist inside gst_face_blur_set_property
function loads the new XML file in the CvHaarClassifierCascade property
without first checking that it is released because maybe there is an XML
file previously loaded.

https://bugzilla.gnome.org/show_bug.cgi?id=753994

9 years agofaceblur: Code refactoring of gst_face_blur_load_profile.
Vanessa Chipirrás Navalón [Wed, 26 Aug 2015 10:27:15 +0000 (12:27 +0200)]
faceblur: Code refactoring of gst_face_blur_load_profile.

Changes inside the gst_face_blur_load_profile function, the number of
input parameters and in lines where it is used due to it cannot be used
generically.

https://bugzilla.gnome.org/show_bug.cgi?id=753994

9 years agofaceblur: Change gstfaceblur to C++.
Vanessa Chipirrás Navalón [Wed, 26 Aug 2015 09:57:24 +0000 (11:57 +0200)]
faceblur: Change gstfaceblur to C++.

Change the gstfaceblur.c file to cpp and add it into Makefile.
It is necessary to migrate the faceblur plugin to C++,
in order to load new and old classifiers, to make faceblur work
with newer versions of Opencv.

https://bugzilla.gnome.org/show_bug.cgi?id=753994

9 years agodvbsuboverlay: Fix caps memory leak by making static caps actually static
Sebastian Dröge [Thu, 27 Aug 2015 08:10:05 +0000 (11:10 +0300)]
dvbsuboverlay: Fix caps memory leak by making static caps actually static

See https://bugzilla.gnome.org/show_bug.cgi?id=754157

9 years agodashdemux: don't meddle with the class struct from an instance
Tim-Philipp Müller [Fri, 21 Aug 2015 15:13:43 +0000 (16:13 +0100)]
dashdemux: don't meddle with the class struct from an instance

Fix some very dubious code. The class methods should always
be set, and the instance-specific check should then be done
inside the method. For data_received that's there already, for
finish_fragment we need to add it.

https://bugzilla.gnome.org/show_bug.cgi?id=753937

9 years agoh264parse/h265parse: Fix negotiation crash
Nicolas Dufresne [Wed, 26 Aug 2015 14:52:46 +0000 (10:52 -0400)]
h264parse/h265parse: Fix negotiation crash

As it's recursive, gst_pad_get_allowed_caps() may also return
empty for anything incompatible downstream. EMPTY is not valid caps
value for gst_caps_fixate(). This lead to assertion and then crash.
Ideally, the negotiate function should be re-factored to have a return
value, and we could make the negotiation fails earlier.

https://bugzilla.gnome.org/show_bug.cgi?id=754122

9 years agocompositor: Actually use the output resolution for clamping
Nirbheek Chauhan [Wed, 26 Aug 2015 10:10:16 +0000 (15:40 +0530)]
compositor: Actually use the output resolution for clamping

The obscured check in compositor was using the dimensions of the pad to clamp
the h/w of the pad instead of the output resolution, and was doing an incorrect
calculation to do so. Fix that by simplifying the whole calculation by using
corner coordinates. Also add a test for this bug which fell through the cracks,
and just skip all the obscured tests if the pad's alpha is 0.0.

https://bugzilla.gnome.org/show_bug.cgi?id=754107

9 years agocodecparsers: h265: Fix tile row and column parsing
XuGuangxin [Mon, 24 Aug 2015 04:46:27 +0000 (07:46 +0300)]
codecparsers: h265: Fix tile row and column parsing

Section 6.5.1:  Coding tree block raster and tile scanning conversion process
Follow the equations 6-3 and 6-4

This will provide correct offset_max in slice_header for parsing
num_entry_point_offsets.

https://bugzilla.gnome.org/show_bug.cgi?id=754024

Signed-off-by: Sreerenj Balachandran <sreerenj.balachandran@intel.com>
9 years agohlsdemux: select correct position for live streams that don't remove fragments
Athanasios Oikonomou [Fri, 7 Aug 2015 09:53:23 +0000 (12:53 +0300)]
hlsdemux: select correct position for live streams that don't remove fragments

Some live streams (eg youtube) don't remove fragments in order to allow
seeking back in time (live + vod).

When gst_m3u8_client_has_next_fragment is called, we are getting wrong fragment
because current_file points in first file of the fragments list resulting in
watching the stream from the beginning again.

This patch sets current_file to nth fragment for live streams, then on
gst_m3u8_client_has_next_fragment will keep up with the live stream.

https://bugzilla.gnome.org/show_bug.cgi?id=753344

9 years agoglupload: Use base class metadata copy function
Nicolas Dufresne [Mon, 24 Aug 2015 23:47:01 +0000 (19:47 -0400)]
glupload: Use base class metadata copy function

This allow properly copying selected meta, like the composition
overlay. Note that output buffer need to be readable, but GlUpload
keeps a ref. For now, simply drop GlUpload ref after perform,
leaving that ref has no purpose. The method shall be removed
in the future.

https://bugzilla.gnome.org/show_bug.cgi?id=754047

9 years agoglcolorconvert: Use base transform metadata copy
Nicolas Dufresne [Mon, 24 Aug 2015 23:28:10 +0000 (19:28 -0400)]
glcolorconvert: Use base transform metadata copy

Use base class default method instead of only copying flags and
timestamp. This way, selected meta's like compostion overlay will
be passed downstream as expected.

https://bugzilla.gnome.org/show_bug.cgi?id=754047

9 years agogl/eagl: Unref context after setting a window handle
Sebastian Dröge [Tue, 25 Aug 2015 07:09:14 +0000 (10:09 +0300)]
gl/eagl: Unref context after setting a window handle

gst_gl_window_get_context() returns a new reference.

Hopefully fixes https://bugzilla.gnome.org/show_bug.cgi?id=753758

9 years agowin32: Add new h265 parser symbols
Sebastian Dröge [Mon, 24 Aug 2015 18:28:46 +0000 (21:28 +0300)]
win32: Add new h265 parser symbols

9 years agocodecparsers: h265: Add APIs for up-right-diagonal/raster scan conversion
lyb [Mon, 24 Aug 2015 05:06:15 +0000 (08:06 +0300)]
codecparsers: h265: Add APIs for up-right-diagonal/raster scan conversion

As per  7-42 and 7-43 the ScalingFactor's scanIdx is 0,
which is "up-right-diagonal" scan. Add APIs for converting
up-right-diagonal to raster and vise versa.

https://bugzilla.gnome.org/show_bug.cgi?id=754024

9 years agocodecparsers: h265: Fix the range of delta_chroma_log2_weight_denom
XuGuangxin [Mon, 24 Aug 2015 01:18:52 +0000 (04:18 +0300)]
codecparsers: h265: Fix the range of delta_chroma_log2_weight_denom

Being more strict on specification, According to 7.4.7.3,
delta_chroma_log2_weight_denom should be in the range of
[(0 - luma_log2_weight_denom), (7 - luma_log2_weight_denom)]

https://bugzilla.gnome.org/show_bug.cgi?id=754024

9 years agotsdemux: fix latency handling again
Tim-Philipp Müller [Sun, 23 Aug 2015 22:38:21 +0000 (23:38 +0100)]
tsdemux: fix latency handling again

The tsdemux latency should always be added to the minimum
latency (which is always a valid clock time value). The
"cleanup" in commit a1f709c2 made it so that it would not
be added if upstream reported 0 as minimum latency (as
e.g. udpsrc would). This broke playback of live mpeg-ts
streaming in some cases, leading to playback stutter due
to a too-small configured latency for the pipeline.

https://bugzilla.gnome.org/show_bug.cgi?id=751508

9 years agoglimagesink: update display size before sending event
Nicolas Dufresne [Fri, 21 Aug 2015 20:42:18 +0000 (13:42 -0700)]
glimagesink: update display size before sending event

This is minor issue, as the reconfigure event is asynchronous.
Basically, update width/height before sending the event.

9 years agogl: Let base transform relay the meta api for us
Nicolas Dufresne [Fri, 21 Aug 2015 00:27:34 +0000 (17:27 -0700)]
gl: Let base transform relay the meta api for us

During allocation query, when this element is not passthrough, it must
relay the overlay compostion meta and it's parameters. Fortunatly, base
transform can do this for us.

https://bugzilla.gnome.org/show_bug.cgi?id=753850

9 years agocurl: Give netinet/ip.h it own configure check
Koop Mast [Fri, 21 Aug 2015 18:29:24 +0000 (20:29 +0200)]
curl: Give netinet/ip.h it own configure check

On FreeBSD netinet/ip.h needs the sys/types.h and netinet/in.h header
before it can be tested.

https://bugzilla.gnome.org/show_bug.cgi?id=753944

9 years agodtsdec: remove unused value
Luis de Bethencourt [Wed, 19 Aug 2015 16:42:49 +0000 (17:42 +0100)]
dtsdec: remove unused value

length is set to a different value before being read. Remove initial value.
Silences compiler warning.

9 years agofacedetect: remove unnecessary variable.
Vanessa Chipirrás Navalón [Fri, 21 Aug 2015 03:26:25 +0000 (05:26 +0200)]
facedetect: remove unnecessary variable.

Memory is reserved for this variable and then released without making any
use of it.

https://bugzilla.gnome.org/show_bug.cgi?id=748377

9 years agoglcontext: add specific error message when missing GL_SHADING_LANGUAGE_VERSION
Justin Kim [Fri, 21 Aug 2015 05:07:32 +0000 (14:07 +0900)]
glcontext: add specific error message when missing GL_SHADING_LANGUAGE_VERSION

GL_SHADING_LANGUAGE_VERSION was introduced since ES 2.0, but in some
android emulator doesn't support this feature. To prevent confusion for
developer, the error message need to be more clear.

https://bugzilla.gnome.org/show_bug.cgi?id=753905

9 years agotests: audiomixer: remove duplicated word in comment
hoonhee.lee [Fri, 21 Aug 2015 07:44:43 +0000 (16:44 +0900)]
tests: audiomixer: remove duplicated word in comment

https://bugzilla.gnome.org/show_bug.cgi?id=753915

9 years agocaopengllayersink: Don't chain up to parent's query handling twice for DRAIN query
hoonhee.lee [Fri, 21 Aug 2015 07:31:41 +0000 (16:31 +0900)]
caopengllayersink: Don't chain up to parent's query handling twice for DRAIN query

https://bugzilla.gnome.org/show_bug.cgi?id=753913

9 years agocheck: hlsdemux: make duration unit test robust to floating point rounding
Alex Ashley [Thu, 20 Aug 2015 14:00:02 +0000 (15:00 +0100)]
check: hlsdemux: make duration unit test robust to floating point rounding

The test_playlist_with_doubles_duration() test fails on some platforms
due to rounding errors that occur when m3u8.c converts from the floating
point value in the HLS manifest to a GstClockTime.

Using assert_equals_float() fixes this because this function handles
the rounding error issues by accepting almost equal.

https://bugzilla.gnome.org/show_bug.cgi?id=753881