platform/upstream/gstreamer.git
11 years agodocs: add new sdp methods to docs
Wim Taymans [Wed, 24 Apr 2013 11:57:23 +0000 (13:57 +0200)]
docs: add new sdp methods to docs

11 years agodef: update with new sdp symbols
Wim Taymans [Wed, 24 Apr 2013 11:50:54 +0000 (13:50 +0200)]
def: update with new sdp symbols

11 years agosdp: add more functions to modify the sdp message
Wim Taymans [Tue, 23 Apr 2013 13:23:05 +0000 (15:23 +0200)]
sdp: add more functions to modify the sdp message

Add functions to insert, replace and remove various sdp message fields
and structures.

See: https://bugzilla.gnome.org/show_bug.cgi?id=698558

11 years agostreamsynchronizer: Fix check for belonging to another stream
Sebastian Dröge [Wed, 24 Apr 2013 09:10:34 +0000 (11:10 +0200)]
streamsynchronizer: Fix check for belonging to another stream

https://bugzilla.gnome.org/show_bug.cgi?id=697820

11 years agostreamsynchronizer: Create element with gst_element_factory_make()
Sebastian Dröge [Wed, 24 Apr 2013 09:07:30 +0000 (11:07 +0200)]
streamsynchronizer: Create element with gst_element_factory_make()

Otherwise plugin_init() is not called and initialization is missing.

11 years agotypefindfunctions: fix crash in new MSS typefinder
Tim-Philipp Müller [Tue, 23 Apr 2013 23:08:15 +0000 (00:08 +0100)]
typefindfunctions: fix crash in new MSS typefinder

Fixes icydemux test_first_buf_offset_when_merged_for_typefinding
unit test segfaulting on a NULL pointer.

11 years agodecodebin2: also remove the bytes limit
Wim Taymans [Wed, 6 Mar 2013 15:44:45 +0000 (16:44 +0100)]
decodebin2: also remove the bytes limit

Remove the byte limit for adaptive http streaming. Because some fragments might
be very big, we might need a lot of buffering. I also suspect another problem
where data is actually missing and things go out of sync somehow.

11 years agodecodebin2: update buffer size in multiqueue
Wim Taymans [Mon, 4 Mar 2013 15:19:18 +0000 (16:19 +0100)]
decodebin2: update buffer size in multiqueue

When we disable buffering in the more upstream multiqueue elements,
we need to also update the queue limits. In particular, the max_size_time should
be set to 0 or else we might simply deadlock.

11 years agodecodebin2: only allow 'lower' multiqueues to emit buffering messages
Thiago Santos [Wed, 6 Feb 2013 11:41:19 +0000 (08:41 -0300)]
decodebin2: only allow 'lower' multiqueues to emit buffering messages

When we have a scenario of demuxers linked to demuxers, decodebin2
will create multiqueue at different levels of the pipeline. The problem
is that only the lowest multiqueue's should do the buffering messaging,
as they will handle with the raw streams data.

When all multiqueues are doing buffering, the upper ones can handle
large buffers that easily fill them, moving from 0% to 100% from
buffer to buffer, causing too much buffering messages to be posted.
This hangs the pipeline unnecessarily and might lead to deadlocks.

11 years agodecodebin2: do not handle the next-groups list as if it was a single item
Thiago Santos [Wed, 6 Feb 2013 14:09:52 +0000 (11:09 -0300)]
decodebin2: do not handle the next-groups list as if it was a single item

Decodebin2's chains store a next_groups list that was being handled as
it could only have a single element. This is true for most of the
chaining streams scenarios where streams change not very often.

In more stressfull changing scenarios, like adaptive streams, those
changes can happen very often, and in short time intervals. This could
confuse decodebin2 as this list was always being used as a single
element list.

This patches makes it handle as a real list, using iteration instead
of picking the first element as the correct one always.

11 years agodecodebin2: preserve next groups order
Thiago Santos [Fri, 1 Feb 2013 20:50:36 +0000 (17:50 -0300)]
decodebin2: preserve next groups order

11 years agodecodebin2: still report chain as drained when not 'handled'
Thiago Santos [Wed, 9 Jan 2013 21:39:49 +0000 (18:39 -0300)]
decodebin2: still report chain as drained when not 'handled'

Even if the chain hasn't been 'handled' in this switching round,
report it as drained so upper chains/groups know abou it.

This makes switching happen on upper levels of the groups/chain
trees

11 years agotypefind: add smoothstreaming manifest typefinding
Thiago Santos [Thu, 11 Apr 2013 12:47:51 +0000 (09:47 -0300)]
typefind: add smoothstreaming manifest typefinding

Checks if the received XML is a smoothstreaming manifest
in both UTF8 and UTF16 formats. The check is made for a
SmoothStreamingMedia top level element.

Conflicts:
gst/typefind/gsttypefindfunctions.c

11 years agostreamsynchronizer: Don't consider a stream added for an already running one as ...
Sebastian Dröge [Tue, 23 Apr 2013 11:54:49 +0000 (13:54 +0200)]
streamsynchronizer: Don't consider a stream added for an already running one as "new"

Fixes enabling visualizations after disabling them after they were enabled already.

11 years agostreamsynchronizer: If a stream belongs to an already running stream, don't wait
Sebastian Dröge [Tue, 23 Apr 2013 11:18:45 +0000 (13:18 +0200)]
streamsynchronizer: If a stream belongs to an already running stream, don't wait

This fixes enabling visualizations after the audio stream already started.

https://bugzilla.gnome.org/show_bug.cgi?id=697820

11 years agoAutomatic update of common submodule
Tim-Philipp Müller [Mon, 22 Apr 2013 22:51:08 +0000 (23:51 +0100)]
Automatic update of common submodule

From 3cb3d3c to 5edcd85

11 years agoUpdate disted orc backup files
Tim-Philipp Müller [Sun, 21 Apr 2013 18:02:42 +0000 (19:02 +0100)]
Update disted orc backup files

Generated with 0.4.17 now.

11 years agouridecodebin: don't report 'no uri handler found' if the URI was rejected by a source
Tim-Philipp Müller [Sun, 21 Apr 2013 16:24:55 +0000 (17:24 +0100)]
uridecodebin: don't report 'no uri handler found' if the URI was rejected by a source

If a source element could be created for a URI, but all elements rejected
the URI for some reason, propagate the error from the URI handler instead
of reporting a 'no uri handler found for protocol xyz' error, which is
confusing. Fixes error reporting with dvb:// URIs when the channel config
file could not be found or not be parsed or the channel isn't listed.

https://bugzilla.gnome.org/show_bug.cgi?id=678892

11 years agoadder: Do not try to wait for flush_stop after receiving a segment event
Thibault Saunier [Fri, 19 Apr 2013 20:59:56 +0000 (17:59 -0300)]
adder: Do not try to wait for flush_stop after receiving a segment event

  + Add a simple test

11 years agovolume: skip controlled processing if we have no timestamp
Stefan Sauer [Thu, 18 Apr 2013 05:55:56 +0000 (07:55 +0200)]
volume: skip controlled processing if we have no timestamp

11 years agoplaybin: use _plugin_feature_rank_compare API instead of duplicating the code.
Sreerenj Balachandran [Thu, 18 Apr 2013 09:07:37 +0000 (12:07 +0300)]
playbin: use _plugin_feature_rank_compare API instead of duplicating the code.

11 years agodecodebin: use _plugin_feature_rank_compare API instead of duplicating the code.
Sreerenj Balachandran [Thu, 18 Apr 2013 09:03:29 +0000 (12:03 +0300)]
decodebin: use _plugin_feature_rank_compare API instead of duplicating the code.

11 years agovideoencoder: Simply setcaps function
Sebastian Dröge [Thu, 18 Apr 2013 07:58:09 +0000 (09:58 +0200)]
videoencoder: Simply setcaps function

11 years agoaudioencoder: Ignore caps events if the input caps did not change
Sebastian Dröge [Thu, 18 Apr 2013 07:54:48 +0000 (09:54 +0200)]
audioencoder: Ignore caps events if the input caps did not change

11 years agoaudiodecoder: Ignore caps events if the input caps did not change
Sebastian Dröge [Thu, 18 Apr 2013 07:54:25 +0000 (09:54 +0200)]
audiodecoder: Ignore caps events if the input caps did not change

11 years agooggmux: add more logging and fix the object param for some logging
Stefan Sauer [Wed, 17 Apr 2013 05:51:38 +0000 (07:51 +0200)]
oggmux: add more logging and fix the object param for some logging

Use data->pad instead of pad for 'object'. Reduce indentation depth in a helper
with an early return.

11 years agovideometa: Extend GstVideoGLTextureUploadMeta
Sebastian Dröge [Wed, 17 Apr 2013 08:35:22 +0000 (10:35 +0200)]
videometa: Extend GstVideoGLTextureUploadMeta

https://bugzilla.gnome.org/show_bug.cgi?id=697112

11 years agodocs: add some more audio macros
Tim-Philipp Müller [Wed, 17 Apr 2013 08:23:20 +0000 (09:23 +0100)]
docs: add some more audio macros

11 years agouridecodebin: query bandwidth capability to source element
Philippe Normand [Thu, 28 Mar 2013 13:21:41 +0000 (14:21 +0100)]
uridecodebin: query bandwidth capability to source element

Use a scheduling query to check if the source element has some
bandwidth limitations. If this is the case on-disk buffering might be
used. If the source element doesn't handle the scheduling query then
fallback to checking the URI protocol against the hardcoded list of
protocols known to handle buffering already.

Fixes bug 693484.

11 years agoapp: Don't use $(GST_PLUGIN_LIBTOOLFLAGS) for real libraries
Sebastian Dröge [Tue, 16 Apr 2013 12:09:43 +0000 (14:09 +0200)]
app: Don't use $(GST_PLUGIN_LIBTOOLFLAGS) for real libraries

11 years agogst: Add better support for static plugins
Sebastian Dröge [Wed, 24 Oct 2012 10:10:44 +0000 (12:10 +0200)]
gst: Add better support for static plugins

11 years agosdp: dynamically allocate buffers
Wim Taymans [Mon, 15 Apr 2013 13:37:17 +0000 (15:37 +0200)]
sdp: dynamically allocate buffers

Remove the static maximum buffer size and replace with dynamic allocation of as
much bytes as needed. Also avoids doing large allocations on the stack.

11 years agosdp: add Since markers
Wim Taymans [Mon, 15 Apr 2013 12:25:16 +0000 (14:25 +0200)]
sdp: add Since markers

11 years agosdp: add boxed type for GstSDPMessage
Jose Antonio Santos Cadenas [Fri, 12 Apr 2013 07:35:34 +0000 (09:35 +0200)]
sdp: add boxed type for GstSDPMessage

Also added some tests of this improvement.

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=697808

11 years agoriff: add format=WMV3 for WMV 3 caps
Matej Knopp [Tue, 9 Apr 2013 20:07:15 +0000 (22:07 +0200)]
riff: add format=WMV3 for WMV 3 caps

https://bugzilla.gnome.org/show_bug.cgi?id=697665

11 years agoplaybin: use ascending order for name based sorting of pluginfeatures.
Sreerenj Balachandran [Mon, 15 Apr 2013 09:37:07 +0000 (12:37 +0300)]
playbin: use ascending order for name based sorting of pluginfeatures.

The compare_factories_func() should return negative value
if the rank of both PluginFeatures are equal and the name of
first PluginFeature comes before the second one (== ascending order).

11 years agodecodebin: use ascending order for name based sorting of pluginfeatures.
Sreerenj Balachandran [Mon, 15 Apr 2013 09:20:09 +0000 (12:20 +0300)]
decodebin: use ascending order for name based sorting of pluginfeatures.

The _decode_bin_compare_factories_func() should return negative
value if the rank of both PluginFeatures are equal and the name of
first PluginFeature comes before the second one (== ascending order).

11 years agoaudioringbuffer: Also reset segbase
Sebastian Dröge [Mon, 15 Apr 2013 08:13:14 +0000 (10:13 +0200)]
audioringbuffer: Also reset segbase

11 years agoaudioringbuffer: Reset segdone when releasing audioringbuffer
Paul HENRYS [Wed, 10 Apr 2013 14:38:14 +0000 (16:38 +0200)]
audioringbuffer: Reset segdone when releasing audioringbuffer

https://bugzilla.gnome.org/show_bug.cgi?id=697723

11 years agortpbasedepayload: Ignore caps events if the caps did not change
Tom Greenwood [Wed, 10 Apr 2013 19:45:37 +0000 (20:45 +0100)]
rtpbasedepayload: Ignore caps events if the caps did not change

https://bugzilla.gnome.org/show_bug.cgi?id=697672

11 years agovideoencoder: Ignore caps events if the caps did not change
Sebastian Dröge [Mon, 15 Apr 2013 07:42:22 +0000 (09:42 +0200)]
videoencoder: Ignore caps events if the caps did not change

11 years agovideodecoder: Ignore caps events if the caps did not change
Tom Greenwood [Wed, 10 Apr 2013 18:07:00 +0000 (19:07 +0100)]
videodecoder: Ignore caps events if the caps did not change

https://bugzilla.gnome.org/show_bug.cgi?id=697672

11 years agoogg: Prevent some symbol conflicts between the ogg plugin and schroedinger
Sebastian Dröge [Mon, 15 Apr 2013 06:38:34 +0000 (08:38 +0200)]
ogg: Prevent some symbol conflicts between the ogg plugin and schroedinger

11 years agoAutomatic update of common submodule
Tim-Philipp Müller [Sun, 14 Apr 2013 16:54:43 +0000 (17:54 +0100)]
Automatic update of common submodule

From 2736592 to 3cb3d3c

11 years agoAutomatic update of common submodule
Tim-Philipp Müller [Sun, 14 Apr 2013 16:26:13 +0000 (17:26 +0100)]
Automatic update of common submodule

From aed87ae to 2736592

11 years agoplaybin: fix jpeg passthrough to decoder sinks by marking image/* as video stream
Tim-Philipp Müller [Fri, 12 Apr 2013 10:49:27 +0000 (11:49 +0100)]
playbin: fix jpeg passthrough to decoder sinks by marking image/* as video stream

11 years agotests: avoid assert in adder test if no audio sources work
Jonas Holmberg [Fri, 12 Apr 2013 09:50:53 +0000 (11:50 +0200)]
tests: avoid assert in adder test if no audio sources work

The array of factories should not contain a NULL element at the end
since the number of arguments is determined via G_N_ELEMENTS and the
NULL will be used as an argument to gst_element_factory_make() if
the other sources in the list weren't usable.

11 years agovideometa: gst_buffer_add_meta() can return NULL
Víctor Manuel Jáquez Leal [Thu, 11 Apr 2013 17:25:51 +0000 (13:25 -0400)]
videometa: gst_buffer_add_meta() can return NULL

https://bugzilla.gnome.org/show_bug.cgi?id=697824

11 years agoAutomatic update of common submodule
Stefan Sauer [Tue, 9 Apr 2013 18:59:51 +0000 (20:59 +0200)]
Automatic update of common submodule

From 04c7a1e to aed87ae

11 years agoximagesink: Fix coompiler error without HAVE_XSHM
Alexander Schrab [Tue, 9 Apr 2013 10:27:48 +0000 (12:27 +0200)]
ximagesink: Fix coompiler error without HAVE_XSHM

https://bugzilla.gnome.org/show_bug.cgi?id=697628

11 years agovideoscale: Fix unit test after latest videoconvert changes
Sebastian Dröge [Tue, 9 Apr 2013 12:04:47 +0000 (14:04 +0200)]
videoscale: Fix unit test after latest videoconvert changes

11 years agovideoconvert: Fix unit test after latest videoconvert changes
Sebastian Dröge [Tue, 9 Apr 2013 11:59:34 +0000 (13:59 +0200)]
videoconvert: Fix unit test after latest videoconvert changes

11 years agortsp: Don't use / as path if no path was provided
Thomas Scheuermann [Mon, 8 Apr 2013 07:09:33 +0000 (09:09 +0200)]
rtsp: Don't use / as path if no path was provided

RTSP does not mandate that a non-zero-length path is used and
some devices (e.g. IQinVision IQeye 1080p) requires that a
zero-length path is used.

11 years agovideoscale: Allow passthrough for ANY caps features
Sebastian Dröge [Mon, 8 Apr 2013 06:29:34 +0000 (08:29 +0200)]
videoscale: Allow passthrough for ANY caps features

11 years agovideoconvert: Allow passthrough for ANY caps features
Sebastian Dröge [Mon, 8 Apr 2013 06:15:15 +0000 (08:15 +0200)]
videoconvert: Allow passthrough for ANY caps features

11 years agovideoscale: set reasonable limits on properties
David Schleef [Sat, 6 Apr 2013 20:00:02 +0000 (13:00 -0700)]
videoscale: set reasonable limits on properties

Properties sharpen, sharpness, and envelope are only useful
near their default values.  Decrease ranges to avoid brokenness.

https://bugzilla.gnome.org/show_bug.cgi?id=682171

11 years agovideoscale: set min value to DBL_MIN to avoid a value of 0.0 that would crash
Stefan Sauer [Fri, 5 Apr 2013 20:03:56 +0000 (22:03 +0200)]
videoscale: set min value to DBL_MIN to avoid a value of 0.0 that would crash

11 years agosdp: Set session as permanent (t=0 0) if there is no time information
Jose Antonio Santos Cadenas [Thu, 4 Apr 2013 14:32:45 +0000 (16:32 +0200)]
sdp: Set session as permanent (t=0 0) if there is no time information

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=697277

11 years agoaudiovisualizer: Don't try to sync controller values to invalid timestamp
Olivier Crête [Tue, 11 Sep 2012 23:41:31 +0000 (19:41 -0400)]
audiovisualizer: Don't try to sync controller values to invalid timestamp

11 years agooggdemux: don't push on NOT_LINKED pads
Wim Taymans [Thu, 4 Apr 2013 16:18:54 +0000 (18:18 +0200)]
oggdemux: don't push on NOT_LINKED pads

If our previous flow return was NOT_LINKED, don't try to push on the pads some
more. If we get a RECONFIGURE event on the pad, try to push on it again.

11 years agodecodebin2: forward all sticky events to decodepad
Wim Taymans [Thu, 4 Apr 2013 13:00:52 +0000 (15:00 +0200)]
decodebin2: forward all sticky events to decodepad

Forward all sticky events to the decodepad before exposing the pads. This makes
sure all sticky events are on the exposed pad.

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=696915

11 years agodecodebin: debug sticky events on exposed pads
Wim Taymans [Thu, 4 Apr 2013 12:53:32 +0000 (14:53 +0200)]
decodebin: debug sticky events on exposed pads

11 years agodecodebin2: small cleanup
Wim Taymans [Thu, 4 Apr 2013 12:37:15 +0000 (14:37 +0200)]
decodebin2: small cleanup

Take the event from the probe just once.

11 years agortsp: make local_ip and remote_ip variables
Wim Taymans [Thu, 4 Apr 2013 10:32:24 +0000 (12:32 +0200)]
rtsp: make local_ip and remote_ip variables

Separate local_ip and remote_ip into separate variables for clarity.

11 years agortsp: calculate the local ip address in accept
Wim Taymans [Thu, 4 Apr 2013 10:16:47 +0000 (12:16 +0200)]
rtsp: calculate the local ip address in accept

Calculate the local IP address in the accept call. We need to place this IP
address in the GET reply in the X-Server-IP-Address header so that the client
knows where to send the POST to in case of tunneled RTSP. Before this patch
it used the client IP address, which would make the client send the POST request
to itself and fail.

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=697092

11 years agovideofilter: add caps to pool config
Matej Knopp [Sat, 30 Mar 2013 16:39:39 +0000 (17:39 +0100)]
videofilter: add caps to pool config

Does not cause problems but it fixes a warning in the log.

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=696916

11 years agooggdemux: Allow EOS on timestamp equal to stop/start
Nicolas Dufresne [Sat, 30 Mar 2013 11:00:35 +0000 (11:00 +0000)]
oggdemux: Allow EOS on timestamp equal to stop/start

Changed the check to a current_time equal to the stop will produce
EOS instead of the next one. Also, segment.start can't be NONE, so removing
this check.

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=696899

11 years agosdp: Do not assign -1 to an unsigned variable, use 0 instead
Jose Antonio Santos Cadenas [Tue, 2 Apr 2013 15:09:48 +0000 (17:09 +0200)]
sdp: Do not assign -1 to an unsigned variable, use 0 instead

Fixes https://bugzilla.gnome.org/show_bug.cgi?id=697162

11 years agosdp: If attribute value is emply do not print colon character
Jose Antonio Santos Cadenas [Tue, 2 Apr 2013 15:05:36 +0000 (17:05 +0200)]
sdp: If attribute value is emply do not print colon character

See https://bugzilla.gnome.org/show_bug.cgi?id=697162

11 years agosdp: Do not add email line by default
Jose Antonio Santos Cadenas [Tue, 2 Apr 2013 15:03:38 +0000 (17:03 +0200)]
sdp: Do not add email line by default

As specified in rfc4566 email line is optional

See https://bugzilla.gnome.org/show_bug.cgi?id=697162

11 years agodocs: add since marker for new video API
Tim-Philipp Müller [Tue, 2 Apr 2013 22:40:13 +0000 (23:40 +0100)]
docs: add since marker for new video API

11 years agopbutils: add description for TrueHD audio
Tim-Philipp Müller [Tue, 2 Apr 2013 21:50:25 +0000 (22:50 +0100)]
pbutils: add description for TrueHD audio

11 years agortsprange: use gst_util_gdouble_to_guint64 in get_seconds
David Svensson Fors [Thu, 28 Mar 2013 22:02:11 +0000 (23:02 +0100)]
rtsprange: use gst_util_gdouble_to_guint64 in get_seconds

https://bugzilla.gnome.org/show_bug.cgi?id=696818

11 years agovisualizer: handle non-existant pool in the default allocation query
Matthew Waters [Thu, 14 Mar 2013 00:15:29 +0000 (11:15 +1100)]
visualizer: handle non-existant pool in the default allocation query

gst_query_set_nth_allocation_pool() requires there to be a pool in the
query already. This is not always the case when we get the query from
upstream.  Use gst_query_add_allocation_pool() instead in such case.

https://bugzilla.gnome.org/show_bug.cgi?id=681719

11 years agovideo-format: Add GST_VIDEO_CAPS_MAKE_WITH_FEATURES() macro
Sebastian Dröge [Mon, 1 Apr 2013 08:21:18 +0000 (10:21 +0200)]
video-format: Add GST_VIDEO_CAPS_MAKE_WITH_FEATURES() macro

11 years agovideometa: Add caps feature #defines for video and GL texture upload meta
Sebastian Dröge [Mon, 1 Apr 2013 08:21:00 +0000 (10:21 +0200)]
videometa: Add caps feature #defines for video and GL texture upload meta

11 years agooggdemux: mark keyframes
Wim Taymans [Sun, 31 Mar 2013 16:54:46 +0000 (18:54 +0200)]
oggdemux: mark keyframes

Use the mapper to set the delta-unit flag.

11 years agotests: fix some printf format compiler warnings
Tim-Philipp Müller [Sun, 31 Mar 2013 16:37:01 +0000 (17:37 +0100)]
tests: fix some printf format compiler warnings

11 years agovideodecoder: always send the new buffer to parse functions
Wim Taymans [Sun, 31 Mar 2013 16:29:07 +0000 (18:29 +0200)]
videodecoder: always send the new buffer to parse functions

When we get a new buffer, always call the parse function, even if it is a 0
sized buffer. For theora we need to also decode 0 sized buffers.
Ideally we would like to make theoradec be packetized but that fails currently
because of oggdemux and because of the assumptions that the base class makes.

11 years agotheoradec: handle empty adapter
Wim Taymans [Sun, 31 Mar 2013 16:28:09 +0000 (18:28 +0200)]
theoradec: handle empty adapter

Should not be empty but might be when we actually pass and decode the 0 packets
later.

11 years agotheoraenc: do not reset the encoder when we need a keyframe
Vincent Penquerc'h [Mon, 25 Feb 2013 10:49:19 +0000 (10:49 +0000)]
theoraenc: do not reset the encoder when we need a keyframe

Instead, remember we need a keyframe, and we will force the encoder
to emit one next time we submit a new frame.
Since libtheora does not have an API to request a keyframe, we reset
the max keyframe interval to 1 temporarily.

This has the advantage that the rate control keeps its history,
and that the encoder won't choose different quant tables or
somesuch, thus requiring new streamheaders (although this is
probably only a theoretical possibility). Should also be a
bit faster than resetting the encoder.

https://bugzilla.gnome.org/show_bug.cgi?id=663350

11 years agoaudiodecoder: don't make negative timestamp
Wim Taymans [Sun, 31 Mar 2013 11:46:30 +0000 (13:46 +0200)]
audiodecoder: don't make negative timestamp

Clamp timestamp interpollation to 0 to avoid going negative. This should not
happen, really, but until the interpolation is improved this seems better.

11 years agostreamsynchronizer: update position for reverse
Wim Taymans [Sun, 31 Mar 2013 10:55:33 +0000 (12:55 +0200)]
streamsynchronizer: update position for reverse

When doing reverse playback the positino advances from timestamp_end to
timestamp.

11 years agoencodebin: Add action signal to get pad for a given profile
Edward Hervey [Thu, 28 Mar 2013 14:20:19 +0000 (15:20 +0100)]
encodebin: Add action signal to get pad for a given profile

This allows getting a pad for a specific encoding profile, which can
be useful when there are several stream profiles of the same type.

Also update the encodebin unit tests so that we check that the returned
pad has the right caps.

https://bugzilla.gnome.org/show_bug.cgi?id=689845

11 years agovideoconvert: use one matrix function
Wim Taymans [Sat, 30 Mar 2013 16:20:42 +0000 (17:20 +0100)]
videoconvert: use one matrix function

Use only one matrix function pointer, let the implementation cast the pixels to
the right size.

11 years agovideoconvert: use one temp array for lines
Wim Taymans [Fri, 29 Mar 2013 10:48:05 +0000 (11:48 +0100)]
videoconvert: use one temp array for lines

Use only one temporary array for pixels.

11 years agovideodecoder: forward stream-start immediately
Wim Taymans [Sat, 30 Mar 2013 18:13:47 +0000 (19:13 +0100)]
videodecoder: forward stream-start immediately

11 years agoaudiodecoder: forward stream-start immediately
Wim Taymans [Sat, 30 Mar 2013 18:07:05 +0000 (19:07 +0100)]
audiodecoder: forward stream-start immediately

11 years agodecodebin: Remove GstdecodePads that are not going to be exposed
Sebastian Dröge [Sat, 30 Mar 2013 16:17:04 +0000 (17:17 +0100)]
decodebin: Remove GstdecodePads that are not going to be exposed

This makes sure that they're unlnked and don't cause any errors or
block the pipeline.

11 years agoplaybin: Ignore caps from audio/video sink factories if there are fixed sinks already
Sebastian Dröge [Sat, 30 Mar 2013 11:15:38 +0000 (12:15 +0100)]
playbin: Ignore caps from audio/video sink factories if there are fixed sinks already

11 years agodmabuf: Some code cleanup
Sebastian Dröge [Sat, 30 Mar 2013 11:08:06 +0000 (12:08 +0100)]
dmabuf: Some code cleanup

11 years agodocs: Add some documentation that was missing before
Sebastian Dröge [Sat, 30 Mar 2013 11:00:35 +0000 (12:00 +0100)]
docs: Add some documentation that was missing before

11 years agoplaybin: Handle caps queries from unlinked elements
Sebastian Dröge [Sat, 30 Mar 2013 10:25:22 +0000 (11:25 +0100)]
playbin: Handle caps queries from unlinked elements

Pass them to all possible sinks and the current sinks to
allow elements to chose a more optimal initial caps.

11 years agodecodebin: Add autoplug-query signal to handle queries for yet unconnected elements
Sebastian Dröge [Fri, 29 Mar 2013 17:27:03 +0000 (18:27 +0100)]
decodebin: Add autoplug-query signal to handle queries for yet unconnected elements

This allows playbin to answer the CAPS query with the possible sink
caps for example, and allows decoders to chose more optimal caps.

11 years agodecodebin: Don't add a children to a decode group twice
Sebastian Dröge [Sat, 30 Mar 2013 09:25:45 +0000 (10:25 +0100)]
decodebin: Don't add a children to a decode group twice

This can happen if a demuxer does not provide fixed caps from the
beginning but only sets them later.

11 years agopbutils: Fix the documentation build warning.
Sreerenj Balachandran [Fri, 29 Mar 2013 14:05:12 +0000 (16:05 +0200)]
pbutils: Fix the documentation build warning.

11 years agoxvimagesink: fix debug message printf format compiler warning
Josep Torra [Sat, 30 Mar 2013 08:39:32 +0000 (09:39 +0100)]
xvimagesink: fix debug message printf format compiler warning

11 years agoaudiotestsrc: fix a comment typo from previous commit
Stefan Sauer [Fri, 29 Mar 2013 16:14:09 +0000 (17:14 +0100)]
audiotestsrc: fix a comment typo from previous commit

11 years agovideometa: Add documentation for GstVideoGLTextureUploadMeta
Sebastian Dröge [Fri, 29 Mar 2013 15:55:56 +0000 (16:55 +0100)]
videometa: Add documentation for GstVideoGLTextureUploadMeta

11 years agoaudiotestssrc: truncate the seek pos to the sample and round the time
Stefan Sauer [Fri, 29 Mar 2013 15:44:17 +0000 (16:44 +0100)]
audiotestssrc: truncate the seek pos to the sample and round the time

Before it was done the other way around and that can trigger the assert that
already is in place. This also makes more sense; when seeking to time x, we want
then sample that is <= that pos.