platform/upstream/gstreamer.git
8 years agompdparser: do not dereference profiles is absent
Vincent Penquerc'h [Fri, 11 Sep 2015 09:57:26 +0000 (10:57 +0100)]
mpdparser: do not dereference profiles is absent

These are used to apply restrictions on what the MPD file may
use, so no profile means no restrictions.

Besides, nothing actually uses the profiles (yet) anyway.

https://bugzilla.gnome.org/show_bug.cgi?id=750869

8 years agoadaptivedemux: fixed multithread support
Florin Apostol [Wed, 16 Sep 2015 15:46:29 +0000 (16:46 +0100)]
adaptivedemux: fixed multithread support

https://bugzilla.gnome.org/show_bug.cgi?id=755169

8 years agodashdemux: parser rejects negative values for mediaPresentationDuration
Florin Apostol [Wed, 28 Oct 2015 16:31:39 +0000 (16:31 +0000)]
dashdemux: parser rejects negative values for mediaPresentationDuration

https://bugzilla.gnome.org/show_bug.cgi?id=752326

8 years agodashdemux: corrected parsing of negative values into unsigned data
Florin Apostol [Wed, 28 Oct 2015 16:24:01 +0000 (16:24 +0000)]
dashdemux: corrected parsing of negative values into unsigned data

https://bugzilla.gnome.org/show_bug.cgi?id=752429

8 years agocodecparser: h264: initialize parsing structures
Víctor Manuel Jáquez Leal [Thu, 17 Sep 2015 18:18:52 +0000 (20:18 +0200)]
codecparser: h264: initialize parsing structures

Initialize to 0 these parse structures before filling them: GstH264SEIMessage,
GstH264NalUnit, GstH264PPS, GstH264SPS and GstH264SliceHdr.

When calling the functions which fill those structures, they may fail, leaving
unitialized those structures. This situation may lead to future problems, such
as a segmentation fault when freeing, for example.

This patch initializes to zero these structures, before filling them.

https://bugzilla.gnome.org/show_bug.cgi?id=755161

8 years agocodecparser: h265: initialize parsing structures
Víctor Manuel Jáquez Leal [Fri, 18 Sep 2015 08:09:26 +0000 (10:09 +0200)]
codecparser: h265: initialize parsing structures

Initialize to 0 these parse structures before filling them: GstH265SEIMessage,
GstH265NalUnit, GstH265VPS, GstH265PPS, GstH265SPS and GstH265SliceHdr.

When calling the functions which fill those structures, they may fail, leaving
unitialized those structures. This situation may lead to future problems, such
as a segmentation fault when freeing, for example.

This patch initializes to zero these structures, before filling them.

https://bugzilla.gnome.org/show_bug.cgi?id=755161

8 years agoliveadder: Make latency property be a uint in millisecs
Olivier Crête [Mon, 26 Oct 2015 23:58:04 +0000 (19:58 -0400)]
liveadder: Make latency property be a uint in millisecs

This restores roughly the same behaviour as the old liveadder element.
Except that the latency now also includes the output-buffer-duration.

https://bugzilla.gnome.org/show_bug.cgi?id=757050

8 years agompdparser: forbid negative values for duration
Vincent Penquerc'h [Tue, 8 Sep 2015 13:00:54 +0000 (14:00 +0100)]
mpdparser: forbid negative values for duration

https://bugzilla.gnome.org/show_bug.cgi?id=752492

8 years agodashdemux: parser rejects XMLs with negative period duration
Florin Apostol [Wed, 28 Oct 2015 15:39:07 +0000 (15:39 +0000)]
dashdemux: parser rejects XMLs with negative period duration

https://bugzilla.gnome.org/show_bug.cgi?id=752329

8 years agompdparser: check segment lists have either duration or timeline
Vincent Penquerc'h [Wed, 28 Oct 2015 15:34:29 +0000 (15:34 +0000)]
mpdparser: check segment lists have either duration or timeline

And add error checking along the way.

Add duration where appropriate so unit tests still pass.

https://bugzilla.gnome.org/show_bug.cgi?id=751650

8 years agompdparser: validate representation set identifier
Vincent Penquerc'h [Tue, 29 Sep 2015 15:17:03 +0000 (16:17 +0100)]
mpdparser: validate representation set identifier

It must have no whitespace, and must comply with RFC 1738 when
used to build a URL.

https://bugzilla.gnome.org/show_bug.cgi?id=750852

8 years agoglslstage: initialize output params to 0 before calling gl functions
Julien Isorce [Wed, 28 Oct 2015 14:41:54 +0000 (14:41 +0000)]
glslstage: initialize output params to 0 before calling gl functions

The client side API of the Chromium's GPU Process has asserts
in debug mode that check that output params are initialized to 0.

8 years agomxfmux: Remove some dead code that could never be called
Sebastian Dröge [Tue, 27 Oct 2015 14:32:48 +0000 (16:32 +0200)]
mxfmux: Remove some dead code that could never be called

Coverity CIDs #1328818, #1328819, #1328820.

8 years agoopengl: build/provide glstereomix/split on gles only systems
Matthew Waters [Mon, 26 Oct 2015 13:36:41 +0000 (00:36 +1100)]
opengl: build/provide glstereomix/split on gles only systems

8 years agogeometrictransform: rename gemetric math functions to have their symbols namespaced
George Kiagiadakis [Mon, 26 Oct 2015 15:51:06 +0000 (16:51 +0100)]
geometrictransform: rename gemetric math functions to have their symbols namespaced

Otherwise those symbols can conflict with external libraries when
linking everything statically for mobile targets.

Use the gst_gm_ prefix, short for gst geometric math.

https://bugzilla.gnome.org/show_bug.cgi?id=756882

8 years agowaylandsink: don't advertise the local wl_display context
George Kiagiadakis [Mon, 26 Oct 2015 14:30:08 +0000 (15:30 +0100)]
waylandsink: don't advertise the local wl_display context

This wl_display proxy is temporary only until waylandsink goes NULL,
at which point the connection to the display is disposed. Unfortunately,
if this is advertised as a GstContext, playbin will cache it and re-feed
it to the sink when it goes PLAYING again, but the wl_display pointer
will at that point be invalid and cause a crash.

Another solution to the problem would be to also cache the GstWlDisplay
object inside the GstContext, which would automatically ref-count
the display connection, but I see no reason in doing that at the moment,
as there are no known users of this GstContext outside waylandsink.
It's probably better to avoid chasing hidden refcounts.

https://bugzilla.gnome.org/show_bug.cgi?id=756567

8 years agoglcolorconvert: allow outputting multiple textures for gles3 contexts
Matthew Waters [Mon, 26 Oct 2015 07:26:18 +0000 (18:26 +1100)]
glcolorconvert: allow outputting multiple textures for gles3 contexts

the USING_GLES2 includes all GLES3 contexts as well which does support
drawing to multiple buffers.  Instead make or decision solely based on
whether glDrawBuffers is available or not.

8 years agoglviewconvert: expose the element on gles2 platforms
Matthew Waters [Mon, 26 Oct 2015 06:36:21 +0000 (17:36 +1100)]
glviewconvert: expose the element on gles2 platforms

We can do everything with gles3 however gles2 restricts us not allowing
separated or frame-by-frame multiview modes due to multiple draw buffers.

8 years agotests: hlsdemux_m3u: add test for master playlist with missing variant list uri
Tim-Philipp Müller [Sun, 25 Oct 2015 00:19:33 +0000 (01:19 +0100)]
tests: hlsdemux_m3u: add test for master playlist with missing variant list uri

https://bugzilla.gnome.org/show_bug.cgi?id=756861

8 years agohlsdemux: don't crash or leak memory on broken master playlist
André Draszik [Tue, 20 Oct 2015 08:13:04 +0000 (09:13 +0100)]
hlsdemux: don't crash or leak memory on broken master playlist

If a (master) playlist contains a variant list entry without a
URI then during parsing of the next variant list entry we are
a) leaking the entry we're currently parsing (new_list), and
b) free'ing the pointer to the previous list entry (list) without
   updating the pointer.

Hence when then adding the URI for the latest parsed entry, incorrect
information is stored, as the information is used from 'list' which
is not valid memory anymore, also leading to crashes.

Fix this by correctly storing the new variant list entry pointer
as needed.

https://bugzilla.gnome.org/show_bug.cgi?id=756861

8 years agohlsdemux: change dispose function to finalize
Tim-Philipp Müller [Fri, 4 Sep 2015 10:29:08 +0000 (11:29 +0100)]
hlsdemux: change dispose function to finalize

8 years agohlsdemux: remove helper function only used once
Tim-Philipp Müller [Sun, 30 Aug 2015 20:35:02 +0000 (21:35 +0100)]
hlsdemux: remove helper function only used once

8 years agohlsdemux: remove deprecated 'fragments-cache' property
Tim-Philipp Müller [Sun, 30 Aug 2015 16:41:23 +0000 (17:41 +0100)]
hlsdemux: remove deprecated 'fragments-cache' property

It's been non-functional for two cycles now, let's
get rid of it.

8 years agohlsdemux: m3u8: move file lookup by sequence directly into code
Tim-Philipp Müller [Wed, 2 Sep 2015 15:40:17 +0000 (16:40 +0100)]
hlsdemux: m3u8: move file lookup by sequence directly into code

Nicer to read, two lines of code less, and also the callback
function should've been a GCompareFunc that returns a gint
and not a boolean (it did work correctly, was just confusing).

8 years agohlsdemux: m3u8: clean-up: remove unused argument in function call
Tim-Philipp Müller [Tue, 1 Sep 2015 11:48:38 +0000 (12:48 +0100)]
hlsdemux: m3u8: clean-up: remove unused argument in function call

8 years agohlsdemux: m3u8: clean-up: reindent header for better readability
Tim-Philipp Müller [Sun, 30 Aug 2015 16:22:19 +0000 (17:22 +0100)]
hlsdemux: m3u8: clean-up: reindent header for better readability

And remove duplicate define.

8 years agomxfmux: Remove empty set/get_property()
Sebastian Dröge [Fri, 23 Oct 2015 15:58:41 +0000 (18:58 +0300)]
mxfmux: Remove empty set/get_property()

We can add them again if we ever add properties.

8 years agomxfmux: Port to GstAggregator
Sebastian Dröge [Fri, 23 Oct 2015 15:38:33 +0000 (18:38 +0300)]
mxfmux: Port to GstAggregator

8 years agomxfmux: Set GC essence element UL version to 0x01
Sebastian Dröge [Thu, 22 Oct 2015 16:20:24 +0000 (19:20 +0300)]
mxfmux: Set GC essence element UL version to 0x01

ffmpeg otherwise rejects it and the spec is not 100% clear about that.

8 years agosrtp: Fix critical warning trying to retrieve SSRC from a non fb message
Santiago Carot-Nemesio [Tue, 20 Oct 2015 15:09:22 +0000 (17:09 +0200)]
srtp: Fix critical warning trying to retrieve SSRC from a non fb message

https://bugzilla.gnome.org/show_bug.cgi?id=756876

8 years agomxf: Reorder CFLAGS and LIBS
Sebastian Dröge [Thu, 22 Oct 2015 16:24:34 +0000 (19:24 +0300)]
mxf: Reorder CFLAGS and LIBS

8 years agoaggregator: Add create_new_pad() vfunc to allow subclasses to override the default...
Sebastian Dröge [Fri, 23 Oct 2015 12:42:24 +0000 (15:42 +0300)]
aggregator: Add create_new_pad() vfunc to allow subclasses to override the default behaviour

Not all aggregator subclasses will have a single pad template called sink_%u
and might do something special depending on what the application requests.

https://bugzilla.gnome.org/show_bug.cgi?id=757018

8 years agomxfmux: Set KAG size to 1
Sebastian Dröge [Wed, 21 Oct 2015 16:00:41 +0000 (19:00 +0300)]
mxfmux: Set KAG size to 1

We're not aligning our output in any way, and 0 is invalid.

8 years agomxfmux: Don't let the header partition's prev_partition point forward to the footer
Sebastian Dröge [Wed, 21 Oct 2015 15:58:06 +0000 (18:58 +0300)]
mxfmux: Don't let the header partition's prev_partition point forward to the footer

8 years agomxftypes: Fix generation of operational pattern UL
Sebastian Dröge [Wed, 21 Oct 2015 15:32:09 +0000 (18:32 +0300)]
mxftypes: Fix generation of operational pattern UL

8 years agomxfdemux-structure: Sync fakesink state with the bin to make the application work...
Sebastian Dröge [Wed, 21 Oct 2015 15:31:43 +0000 (18:31 +0300)]
mxfdemux-structure: Sync fakesink state with the bin to make the application work at all

8 years agomxfmpeg: h264 is always byte-stream inside MXF
Sebastian Dröge [Wed, 21 Oct 2015 15:20:16 +0000 (18:20 +0300)]
mxfmpeg: h264 is always byte-stream inside MXF

8 years agomxf: Add unit tests for H264/PCM
Sebastian Dröge [Wed, 21 Oct 2015 15:06:11 +0000 (18:06 +0300)]
mxf: Add unit tests for H264/PCM

8 years agomxfmux: Don't leak SEEK events when dropping them
Sebastian Dröge [Wed, 21 Oct 2015 14:47:00 +0000 (17:47 +0300)]
mxfmux: Don't leak SEEK events when dropping them

8 years agomxfvc3: The wrapping is the 15th byte of the essence container UL, not the 16th
Sebastian Dröge [Wed, 21 Oct 2015 14:21:09 +0000 (17:21 +0300)]
mxfvc3: The wrapping is the 15th byte of the essence container UL, not the 16th

In other mappings it is the 16th though.

8 years agomxfdemux: If seeking to the remaining parts of the file fails on EOS, consider the...
Sebastian Dröge [Wed, 21 Oct 2015 14:11:03 +0000 (17:11 +0300)]
mxfdemux: If seeking to the remaining parts of the file fails on EOS, consider the stream done

Without this we would run this while loop forever, always seeking again for
the same stream.

8 years agomxf: Properly enable unit tests again
Sebastian Dröge [Wed, 21 Oct 2015 12:07:23 +0000 (15:07 +0300)]
mxf: Properly enable unit tests again

8 years agomxf: Re-enable mxfmux element
Sebastian Dröge [Tue, 20 Oct 2015 22:22:11 +0000 (01:22 +0300)]
mxf: Re-enable mxfmux element

8 years agomxfjpeg2000: Fix caps
Sebastian Dröge [Tue, 20 Oct 2015 22:21:37 +0000 (01:21 +0300)]
mxfjpeg2000: Fix caps

There is no fields field anymore.

8 years agomxfmux: Don't fail SEGMENT events
Sebastian Dröge [Tue, 20 Oct 2015 22:17:04 +0000 (01:17 +0300)]
mxfmux: Don't fail SEGMENT events

8 years agomxfmux: Push stream-start and caps event before segment event
Sebastian Dröge [Tue, 20 Oct 2015 22:02:46 +0000 (01:02 +0300)]
mxfmux: Push stream-start and caps event before segment event

And don't push the caps event during instance initialization already.

8 years agomxfaes-bwf: Format is unsigned if it's 1 byte per channel
Sebastian Dröge [Tue, 20 Oct 2015 21:45:19 +0000 (00:45 +0300)]
mxfaes-bwf: Format is unsigned if it's 1 byte per channel

... not 1 byte per block.

8 years agoopencv: handdetect: no need to get a writable buffer
Luis de Bethencourt [Wed, 21 Oct 2015 15:51:10 +0000 (16:51 +0100)]
opencv: handdetect: no need to get a writable buffer

No need to get a writable buffer inside cv_trans_func of opencv video
filter. GstBaseTransform takes care of this.

8 years agoAutomatic update of common submodule
Tim-Philipp Müller [Wed, 21 Oct 2015 13:35:11 +0000 (14:35 +0100)]
Automatic update of common submodule

From b99800a to b319909

8 years agoUse new GST_ENABLE_EXTRA_CHECKS #define
Sebastian Dröge [Tue, 20 Oct 2015 14:29:42 +0000 (17:29 +0300)]
Use new GST_ENABLE_EXTRA_CHECKS #define

https://bugzilla.gnome.org/show_bug.cgi?id=756870

8 years agoAutomatic update of common submodule
Sebastian Dröge [Wed, 21 Oct 2015 11:26:01 +0000 (14:26 +0300)]
Automatic update of common submodule

From 9aed1d7 to b99800a

8 years agoglshaderelement: free the vertex and fragment strings on finalize
Matthew Waters [Tue, 20 Oct 2015 18:34:18 +0000 (05:34 +1100)]
glshaderelement: free the vertex and fragment strings on finalize

Fixes a memory leak when using the vertex/fragment properties.

8 years agoandroidmedia: fix build errors
Matthew Waters [Tue, 20 Oct 2015 18:33:13 +0000 (05:33 +1100)]
androidmedia: fix build errors

printf %lld instead of G_GINT64_FORMAT.
Double ret variable declaration.

8 years agoglshaderelement: remove unneeded reference to <gst/gl/gstglshadervariables.h>
Matthew Waters [Tue, 20 Oct 2015 18:00:11 +0000 (05:00 +1100)]
glshaderelement: remove unneeded reference to <gst/gl/gstglshadervariables.h>

8 years agogl: fix usage of string.h and stdio.h functions without including the headers
Matthew Waters [Tue, 20 Oct 2015 17:44:50 +0000 (04:44 +1100)]
gl: fix usage of string.h and stdio.h functions without including the headers

e.g:
gstglcontext_egl.c:613:7: error: implicit declaration of function 'strcmp'
[-Werror=implicit-function-declaration]
   if (strcmp (G_MODULE_SUFFIX, "so") == 0)

8 years agoglimagesink: create a context in NULL_READY
Matthew Waters [Thu, 15 Oct 2015 20:03:06 +0000 (07:03 +1100)]
glimagesink: create a context in NULL_READY

So that it's possible for decoders et al. to request the OpenGL context
in their READY_PAUSED transition with decodebin/playbin.

8 years agoamcviddec: use gstcontext to retreive the OpenGL context
Matthew Waters [Thu, 15 Oct 2015 13:34:22 +0000 (00:34 +1100)]
amcviddec: use gstcontext to retreive the OpenGL context

8 years agoandroidmedia: Only allow GL output if the decoder has unknown color formats
Matthieu Bouron [Mon, 20 Apr 2015 11:46:58 +0000 (13:46 +0200)]
androidmedia: Only allow GL output if the decoder has unknown color formats

If GST_AMC_IGNORE_UNKNOWN_COLOR_FORMATS is set to yes, non-GL output
is still allowed.

https://bugzilla.gnome.org/show_bug.cgi?id=731204

8 years agoandroidmedia: Add support for GL output in amcvideodec
Matthieu Bouron [Thu, 5 Jun 2014 08:33:56 +0000 (10:33 +0200)]
androidmedia: Add support for GL output in amcvideodec

https://bugzilla.gnome.org/show_bug.cgi?id=731204

8 years agoandroidmedia: Do not flush codec if it is not started
Matthieu Bouron [Thu, 30 Apr 2015 10:33:58 +0000 (12:33 +0200)]
androidmedia: Do not flush codec if it is not started

8 years agoandroidmedia: Allow object to be NULL in gst_amc_jni_*unref functions
Matthieu Bouron [Mon, 13 Apr 2015 11:10:10 +0000 (13:10 +0200)]
androidmedia: Allow object to be NULL in gst_amc_jni_*unref functions

8 years agoandroidmedia: Fix debug statement (%d for a gsize argument)
Edward Hervey [Fri, 13 Mar 2015 15:13:08 +0000 (16:13 +0100)]
androidmedia: Fix debug statement (%d for a gsize argument)

8 years agoandroidmedia: Improve color format debug messages
Matthieu Bouron [Thu, 2 Apr 2015 14:28:14 +0000 (16:28 +0200)]
androidmedia: Improve color format debug messages

8 years agoandroidmedia: Improve debug messages
Matthieu Bouron [Tue, 31 Mar 2015 14:24:40 +0000 (16:24 +0200)]
androidmedia: Improve debug messages

8 years agoandroidmedia: Do not warn we do not support COLOR_FormatSurface
Matthieu Bouron [Tue, 31 Mar 2015 15:48:59 +0000 (17:48 +0200)]
androidmedia: Do not warn we do not support COLOR_FormatSurface

8 years agogl: remove unneeded shader variable parsing code
Matthew Waters [Tue, 20 Oct 2015 16:59:33 +0000 (03:59 +1100)]
gl: remove unneeded shader variable parsing code

8 years agoglshaderelement: implement setting arbitrary uniforms
Matthew Waters [Tue, 20 Oct 2015 16:23:30 +0000 (03:23 +1100)]
glshaderelement: implement setting arbitrary uniforms

Currently float and int are supported by default. vec2, vec3, vec4
and mat4 are supported if graphene is used.  Of course if one wants
to set custom uniforms they can also be set using the create-shader
signal.

8 years agoopencv: handdetect: free best_r with delete
Luis de Bethencourt [Tue, 20 Oct 2015 11:24:11 +0000 (12:24 +0100)]
opencv: handdetect: free best_r with delete

best_r is a cv::Rect, so it should be freed with delete and not with
g_free()

8 years agoopencv: motioncells: remove unnecessary check
Luis de Bethencourt [Tue, 20 Oct 2015 11:08:59 +0000 (12:08 +0100)]
opencv: motioncells: remove unnecessary check

We know that the gchar arrays contain at least one string. Furthermore,
g_strfreev() checks if the array is NULL and simply returns if it is.
Hence, there is no need to check if the array is empty before using
g_strfreev().

CID 1327412-1327415

8 years agohls/m3u8: Update current position in all cases
Edward Hervey [Wed, 8 Jul 2015 15:17:12 +0000 (17:17 +0200)]
hls/m3u8: Update current position in all cases

In order to ensure the sequence_position will always be consistently updated,
store the current file duration.

This way, when we advance, we can always increment the position based on what
was previously outputted.

https://bugzilla.gnome.org/show_bug.cgi?id=752132

8 years agodshowvideosink: Fix 64bit compatibility issue
plamot [Wed, 14 Oct 2015 16:35:00 +0000 (18:35 +0200)]
dshowvideosink: Fix 64bit compatibility issue

SetWindowLong works only for 32 bit systems, thus windows events (move/resize)
where not interpreted on 64 bit systems

https://bugzilla.gnome.org/show_bug.cgi?id=756617

8 years agogl: be consistent in gobject boilerpate
Matthew Waters [Mon, 19 Oct 2015 04:15:30 +0000 (15:15 +1100)]
gl: be consistent in gobject boilerpate

GST_GL_IS_* vs GST_IS_GL_*

git grep -l 'GST_GL_IS_' | xargs sed -i 's/GST_GL_IS_/GST_IS_GL_/g'

8 years agoadd glliveshader program to .gitignore
Matthew Waters [Mon, 19 Oct 2015 04:02:10 +0000 (15:02 +1100)]
add glliveshader program to .gitignore

8 years agopnmdec: drop frame in case of _handle() failure
Reynaldo H. Verdejo Pinochet [Sat, 17 Oct 2015 19:48:11 +0000 (12:48 -0700)]
pnmdec: drop frame in case of _handle() failure

Allows baseclass to handle it from there

Related to:

https://bugzilla.gnome.org/show_bug.cgi?id=756563

8 years agopnmdec: propagate input state after parsing
Reynaldo H. Verdejo Pinochet [Sat, 17 Oct 2015 08:51:24 +0000 (01:51 -0700)]
pnmdec: propagate input state after parsing

Store and copy input state fields when setting the
output state of the decoder. Avoids problems like
the framerate set by an upstream element being ignored

Related to:

https://bugzilla.gnome.org/show_bug.cgi?id=756563

8 years agopnmdec: completely reset parsing state at flush
Reynaldo H. Verdejo Pinochet [Sat, 17 Oct 2015 03:45:42 +0000 (20:45 -0700)]
pnmdec: completely reset parsing state at flush

Makes sure the mngr struct reflects a clean state
for the next frame, avoiding failures like:

https://bugzilla.gnome.org/show_bug.cgi?id=756563

8 years agoglshaderelement: implement on-demand create-shader signalling
Matthew Waters [Sat, 17 Oct 2015 04:26:46 +0000 (15:26 +1100)]
glshaderelement: implement on-demand create-shader signalling

One may not have an GstGLContext available or current in the thread where one
would need to update the shader.  Support this by signalling create-shader
whenever the one-shot 'update-shader' is set to TRUE.

8 years agogl/calayer: don't leak a gl shader object
Matthew Waters [Fri, 16 Oct 2015 15:42:47 +0000 (02:42 +1100)]
gl/calayer: don't leak a gl shader object

We will always overwrite the ca_sink->redisplay_shader without freeing the
object.

8 years agogtk: separate out the widget/window destroy callbacks
Matthew Waters [Fri, 16 Oct 2015 15:40:50 +0000 (02:40 +1100)]
gtk: separate out the widget/window destroy callbacks

Fixes assertion due to the sink_finalize() being run before the widget destroy
callback.

https://bugzilla.gnome.org/show_bug.cgi?id=755969

8 years agogl/examples: add a live shader demo using the new GstGLSLStage
Matthew Waters [Fri, 16 Oct 2015 14:08:29 +0000 (01:08 +1100)]
gl/examples: add a live shader demo using the new GstGLSLStage

Implemented with videotestsrc ! glshader ! glupload ! gtkglsink

Errors on an invalid shader compilation are ignored however any error
provided by the glsl compiler is printed to stdout.

8 years agosrtpdec: skip padding when mapping RTP packets
Miguel París Díaz [Thu, 15 Oct 2015 22:23:56 +0000 (00:23 +0200)]
srtpdec: skip padding when mapping RTP packets

https://bugzilla.gnome.org/show_bug.cgi?id=756653

8 years agoamcvideodec: Properly forward the return value of gst_video_decoder_negotiate()
Sebastian Dröge [Wed, 14 Oct 2015 16:32:07 +0000 (19:32 +0300)]
amcvideodec: Properly forward the return value of gst_video_decoder_negotiate()

https://bugzilla.gnome.org/show_bug.cgi?id=756578

8 years agoamcvideodec: Implement support for COLOR_QCOM_FormatYVU420SemiPlanar32mMultiView...
Sebastian Dröge [Wed, 14 Oct 2015 16:24:22 +0000 (19:24 +0300)]
amcvideodec: Implement support for COLOR_QCOM_FormatYVU420SemiPlanar32mMultiView for decoding

https://bugzilla.gnome.org/show_bug.cgi?id=756578

8 years agoglslstage: add brackets around pointer dereference referencing an array
Matthew Waters [Thu, 15 Oct 2015 15:57:19 +0000 (02:57 +1100)]
glslstage: add brackets around pointer dereference referencing an array

While technically, i is always 0 and *vertex_sources[i++] is equivalant
to (*vertex_sources)[i++].  Be future-proof in the case of code
moves/changes/etc.

CID 1327406

8 years agogl/caopengllayersink: port to new GstGLShader API
Matthew Waters [Thu, 15 Oct 2015 12:47:11 +0000 (23:47 +1100)]
gl/caopengllayersink: port to new GstGLShader API

fixes build error:
"undefined symbols for architecture:
gst_gl_shader_compile_with_default_vf_and_check"

8 years agogl/tests: port glcontext test to opengl
Matthew Waters [Thu, 15 Oct 2015 11:42:26 +0000 (22:42 +1100)]
gl/tests: port glcontext test to opengl

Now uses vao's and vbo's when possible like the rest of the gstgl library.

8 years agoglshader: port element to GstGLSLStage
Matthew Waters [Fri, 4 Sep 2015 06:16:51 +0000 (16:16 +1000)]
glshader: port element to GstGLSLStage

- Provide a shader property to set the full shader pipeline
- Provide vertex and fragment properties for just providing simple
  shader sources.

8 years agoglshader: port to using GstGLSLStage objects for string management
Matthew Waters [Fri, 4 Sep 2015 06:02:32 +0000 (16:02 +1000)]
glshader: port to using GstGLSLStage objects for string management

A GstGLShader is now simply a collection of stages that are
compiled and linked together into a program.  The uniform/attribute
interface has remained the same.

8 years agoglsl: add a shader stage object GstGLSLStage
Matthew Waters [Fri, 4 Sep 2015 04:36:47 +0000 (14:36 +1000)]
glsl: add a shader stage object GstGLSLStage

Represents a stage (vertex, geometry, fragment, etc) in the shader
pipeline.

8 years agogl/shader/variables: include generic gl.h header
Matthew Waters [Thu, 3 Sep 2015 14:09:09 +0000 (00:09 +1000)]
gl/shader/variables: include generic gl.h header

Otherwise we may miss some forward declarations

Fixes build error: undefined reference to GstGLSLStage

8 years agogl: add some GLSL utility functions
Matthew Waters [Thu, 20 Aug 2015 05:11:06 +0000 (15:11 +1000)]
gl: add some GLSL utility functions

Specifically parsing/setting GLSL versions and the shader related
function table.

8 years agodtls: Fix name conflict with openssl on win32
Paul Arzelier [Wed, 14 Oct 2015 20:51:29 +0000 (22:51 +0200)]
dtls: Fix name conflict with openssl on win32

https://bugzilla.gnome.org/show_bug.cgi?id=756592

8 years agohls: if media sequence is not specified, use 0
Thiago Santos [Fri, 4 Sep 2015 21:40:18 +0000 (18:40 -0300)]
hls: if media sequence is not specified, use 0

Allows playlists that are missing the mediasequence information to
be correctly parsed. If the playlist was updated without reseting
the mediasequence it would constantly increase over subsequent updates,
leading to issues during playback.

8 years agowaylandsink: fix fallback buffer allocation
George Kiagiadakis [Sun, 30 Aug 2015 12:56:57 +0000 (14:56 +0200)]
waylandsink: fix fallback buffer allocation

Copy-paste mistake; the wl_buffer was attached on the wrong buffer...

8 years agowaylandsink: rename GstWlDisplay::formats array to shm_formats
George Kiagiadakis [Sun, 30 Aug 2015 12:55:11 +0000 (14:55 +0200)]
waylandsink: rename GstWlDisplay::formats array to shm_formats

These formats are specific to the wl_shm interface. We are going
to add dmabuf formats later as well.

8 years agowaylandsink: replace the custom GstWlShmMemory with GstFdMemory
George Kiagiadakis [Sat, 29 Aug 2015 20:29:17 +0000 (22:29 +0200)]
waylandsink: replace the custom GstWlShmMemory with GstFdMemory

8 years agopo: Update POTFILES.in for OpenCV C++ file rename
Sebastian Dröge [Wed, 14 Oct 2015 16:53:34 +0000 (19:53 +0300)]
po: Update POTFILES.in for OpenCV C++ file rename

8 years agohls: Allow debugging of the playlist
Edward Hervey [Wed, 14 Oct 2015 15:41:05 +0000 (17:41 +0200)]
hls: Allow debugging of the playlist

Helps debugging issues

8 years agohlsdemux: Avoid negative sequence numbers
Edward Hervey [Wed, 14 Oct 2015 15:38:39 +0000 (17:38 +0200)]
hlsdemux: Avoid negative sequence numbers

For live streams, we want to make sure there's a certain distance
between the sequence to play and the last (earliest) fragment.

The problem is that it assumes there are at least 3 fragments in
the playlist, which might not always be the case (like in the case
of a server restarting and gradually adding fragments).

In order to avoid ending up with negative sequence numbers (which
will just loop forever), limit the new target sequence number to
the highest of:
* either the first sequence number of the playlist (fallback)
* or 3 fragments from the last one (standard behaviour)

8 years agoopencv: switch remaining to C++
Luis de Bethencourt [Wed, 14 Oct 2015 14:30:05 +0000 (15:30 +0100)]
opencv: switch remaining to C++

Switch remaining C files to C++ for consistency with other OpeCV elements,
and support of the new 2.4.11 API.

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