platform/upstream/gstreamer.git
13 years agoqtmux: Use complete MPEG caps in the unit test
Sebastian Dröge [Wed, 7 Sep 2011 12:26:01 +0000 (14:26 +0200)]
qtmux: Use complete MPEG caps in the unit test

13 years agodocs: cleanup makefiles
Stefan Sauer [Wed, 7 Sep 2011 12:18:58 +0000 (14:18 +0200)]
docs: cleanup makefiles

Remove commented out parts that we don't need. Remove "the wingo addition" - no
so useful after all. Narrow down file-globs for plugin docs.

13 years agosouphttpsrc: Don't handle HTTP response 407 as error if proxy authentication data...
Konstantin Miller [Mon, 29 Aug 2011 12:12:22 +0000 (14:12 +0200)]
souphttpsrc: Don't handle HTTP response 407 as error if proxy authentication data is available

Fixes bug #657422.

13 years agoac3parse: Add Converter to the classification because it can convert between differen...
Sebastian Dröge [Wed, 7 Sep 2011 10:11:39 +0000 (12:11 +0200)]
ac3parse: Add Converter to the classification because it can convert between different alignments

This allows decodebin2 to let it negotiate properly.

13 years agoaudioparsers: Improve src template caps
Sebastian Dröge [Wed, 7 Sep 2011 10:10:48 +0000 (12:10 +0200)]
audioparsers: Improve src template caps

Remove the parsed/framed fields and add all fields to the template
caps that always exist.

13 years agoaacparse: parse codec_data to determine number of samples per frame
Mark Nauwelaerts [Tue, 6 Sep 2011 13:59:49 +0000 (15:59 +0200)]
aacparse: parse codec_data to determine number of samples per frame

Fixes #656734.

13 years agoAutomatic update of common submodule
Stefan Sauer [Tue, 6 Sep 2011 19:24:46 +0000 (21:24 +0200)]
Automatic update of common submodule

From a39eb83 to 11f0cd5

13 years agoAutomatic update of common submodule
Stefan Sauer [Tue, 6 Sep 2011 13:40:32 +0000 (15:40 +0200)]
Automatic update of common submodule

From 605cd9a to a39eb83

13 years agomatroskamux: make default duration check less sensitive
Mark Nauwelaerts [Tue, 6 Sep 2011 13:05:37 +0000 (15:05 +0200)]
matroskamux: make default duration check less sensitive

Frame duration might vary for 1 usecond, in this case matroskamux
decides to create BLOCKGROUP instead of SIMPLEBLOCK.

Convert duration to timecodescale which is (typically) less precise, and
then also allow the difference of 1/-1 to arrange for less sensitive check.

Based on patch by Alexey Fisher <bug-track@fisher-privat.net>

Fixes #653080.

13 years agortpmp4gdepay: improve bogus interleaved index compensating
Mark Nauwelaerts [Tue, 6 Sep 2011 11:18:40 +0000 (13:18 +0200)]
rtpmp4gdepay: improve bogus interleaved index compensating

Patch by <gudake@gmail.com>

Fixes #654585.

13 years agosouphttpsrc: Allow positive, non-1.0 segment rates
Sebastian Dröge [Tue, 6 Sep 2011 08:33:21 +0000 (10:33 +0200)]
souphttpsrc: Allow positive, non-1.0 segment rates

Only negative rates are not supported. Fixes bug #658305.

13 years agotests: parsers: provide more real data when testing draining of garbage
Mark Nauwelaerts [Mon, 5 Sep 2011 13:50:56 +0000 (15:50 +0200)]
tests: parsers: provide more real data when testing draining of garbage

13 years agoamrparse: fix and streamline valid frame checking
Mark Nauwelaerts [Mon, 5 Sep 2011 13:50:04 +0000 (15:50 +0200)]
amrparse: fix and streamline valid frame checking

... to handle various combinations of sync or not, and sufficient data
or not as might be expected.

Fixes #650714.

13 years agoqtdemux: fragmented support; avoid adjustment for keyframe seek
Mark Nauwelaerts [Mon, 5 Sep 2011 12:49:32 +0000 (14:49 +0200)]
qtdemux: fragmented support; avoid adjustment for keyframe seek

... since all index data may not yet be available at that time.

13 years agoqtdemux: fragmented support; mark all audio track samples as keyframe
Mark Nauwelaerts [Mon, 5 Sep 2011 12:48:02 +0000 (14:48 +0200)]
qtdemux: fragmented support; mark all audio track samples as keyframe

13 years agoqtdemux: fragmented support; properly init return variable value
Brian Li [Mon, 5 Sep 2011 12:46:29 +0000 (14:46 +0200)]
qtdemux: fragmented support; properly init return variable value

Fixes #655918.

13 years agortspsrc: add gtk-doc for new short-header property
Mark Nauwelaerts [Mon, 5 Sep 2011 11:31:20 +0000 (13:31 +0200)]
rtspsrc: add gtk-doc for new short-header property

13 years agortspsrc: allow sending short RTSP requests to a server
Marc Leeman [Mon, 5 Sep 2011 11:18:39 +0000 (13:18 +0200)]
rtspsrc: allow sending short RTSP requests to a server

Some encoders (Arecont) do not like the long OPTIONS sent at startup as sent by
GStreamer, but do accept the short header as sent by Live555.

This patch makes the extending the request optional by adding a property
(short-header).

Fixes #655805.

API: GstRTSPSrc:short-header

13 years agortph263ppay: Set H263-2000 if thats what the other side wants
Olivier Crête [Wed, 4 Mar 2009 19:51:09 +0000 (14:51 -0500)]
rtph263ppay: Set H263-2000 if thats what the other side wants

The static caps states this element supports H263-2000, but setcaps never
sets it, so it was lie.

See https://bugzilla.gnome.org/show_bug.cgi?id=577784

13 years agortpsession: Initialise the last_keyframe_request variable
Olivier Crête [Tue, 30 Aug 2011 23:02:51 +0000 (19:02 -0400)]
rtpsession: Initialise the last_keyframe_request variable

13 years agomultiudpsink: make add/remove/clear/get-stats action signals
Peter Korsgaard [Wed, 31 Aug 2011 14:04:24 +0000 (16:04 +0200)]
multiudpsink: make add/remove/clear/get-stats action signals

http://bugzilla.gnome.org/show_bug.cgi?id=657830

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
13 years agoqtdemux: push mode; perform some extra checks prior to upstream seeking
Mark Nauwelaerts [Tue, 30 Aug 2011 11:33:49 +0000 (13:33 +0200)]
qtdemux: push mode; perform some extra checks prior to upstream seeking

13 years agoqtdemux: push mode; fix buffered streaming
Mark Nauwelaerts [Tue, 30 Aug 2011 11:28:21 +0000 (13:28 +0200)]
qtdemux: push mode; fix buffered streaming

That is, in case where no seek is peformed to moov, but preceding
limited mdat is buffered.

13 years agoqtdemux: avoid overflow wraparound in timestamp when adding durations
Mark Nauwelaerts [Mon, 29 Aug 2011 13:13:56 +0000 (15:13 +0200)]
qtdemux: avoid overflow wraparound in timestamp when adding durations

Do some type juggling to avoid overflow, while still allowing for 'negative'
durations (which would need a wraparound effect).

13 years agov4l2src: make this work more than once in a row
Vincent Penquerc'h [Thu, 25 Aug 2011 22:37:47 +0000 (23:37 +0100)]
v4l2src: make this work more than once in a row

We used to skip frame rate setup if the camera was already setup
with the requested frame rate. This breaks some cameras though,
causing them to not output data (several models of Thinkpad cameras
have this problem at least).
So, don't skip.

https://bugzilla.gnome.org/show_bug.cgi?id=638300

13 years agoaacparse: only require two frames in a row when we do not have sync
Vincent Penquerc'h [Tue, 23 Aug 2011 11:12:15 +0000 (12:12 +0100)]
aacparse: only require two frames in a row when we do not have sync

This avoids a single bit error dropping two frames unnecessarily.
The two consecutive frames check is still required when we don't
have sync.

https://bugzilla.gnome.org/show_bug.cgi?id=657080

13 years agopulsesink: Trivial indentation fix
Arun Raghavan [Tue, 23 Aug 2011 16:11:15 +0000 (21:41 +0530)]
pulsesink: Trivial indentation fix

13 years agoflacdec: Correct sample number rounding resulting in timestamp jitter
Monty Montgomery [Thu, 21 Jul 2011 21:23:28 +0000 (17:23 -0400)]
flacdec: Correct sample number rounding resulting in timestamp jitter

flacdec converts the src timestamp to a sample number, uses that internally, then reconverts the sample number to a timestamp for the output buffer.  Unfortunately, sample numbers can't be represented in an integer number of nanoseconds, and the conversion process was truncating rather than rounding, resulting in sample numbers and output timestamps that were often off by a full sample.

This corrects the time->sample convesion

13 years agobreakmydata: element is not passthrough
David Schleef [Sat, 20 Aug 2011 21:48:20 +0000 (14:48 -0700)]
breakmydata: element is not passthrough

13 years agomultifilesrc: quiet debugging
David Schleef [Wed, 13 Jul 2011 18:20:34 +0000 (11:20 -0700)]
multifilesrc: quiet debugging

13 years agodeinterlace: change field handling through methods
David Schleef [Mon, 11 Jul 2011 04:40:20 +0000 (21:40 -0700)]
deinterlace: change field handling through methods

This likely breaks stuff.  The good: all of the methods now create
field images aligned with input frames, without timestamp mangling.
The bad: this touches a lot of code, much of which is hairy and in
need of cleanup.  However, at this point we can reasonably create a
PSNR-based test.

13 years agomultifilesink: reset ->streamheaders to NULL on _stop
Alessandro Decina [Sun, 21 Aug 2011 12:41:14 +0000 (14:41 +0200)]
multifilesink: reset ->streamheaders to NULL on _stop

Fixes invalid memory access reusing multifilesink

13 years agopulsesink: Allow writes in bigger chunks
David Henningsson [Thu, 18 Aug 2011 11:37:39 +0000 (13:37 +0200)]
pulsesink: Allow writes in bigger chunks

There's no use in splitting the incoming data down to the segsize
limit - by writing as much as possible in one chunk, we increase
performance and avoid PulseAudio unnecessary rewinds.

Signed-off-by: David Henningsson <david.henningsson@canonical.com>
13 years agomatroskademux: ensure no-more-pads is always emitted
Vincent Penquerc'h [Mon, 8 Aug 2011 21:14:28 +0000 (22:14 +0100)]
matroskademux: ensure no-more-pads is always emitted

In particular, do so even if failing to read while prerolling,
such as when reading from a partial file (eg, while it is being
downloaded).

This fixes a wedge in playbin2.

https://bugzilla.gnome.org/show_bug.cgi?id=651965

13 years agoflacdec: avoid timestamp/offset tracking going out of sync
Vincent Penquerc'h [Tue, 16 Aug 2011 16:27:13 +0000 (17:27 +0100)]
flacdec: avoid timestamp/offset tracking going out of sync

The libFLAC API is callback based, and we must only call it to
output data when we know we have enough input data. For this
reason, a single processing step is done when receiving a buffer.
However, if there were metadata buffers still pending, a step
intended for the first audio frame might end up writing that
leftover metadata. Since a single step is done per buffer, this
will cause every buffer to be written one step late.

This would add some latency (a bufferfull's worth), possibly
lose a buffer when seeking or the like, and also cause timestamp
and offset to be applied to the wrong buffer, as updates to
the "current" segment last_stop (from incoming buffer timestamp)
will be applied to an output buffer originating from the previous
incoming buffer.

This fixes the issue by ensuring that, upon receiving the first
audio frame, processing is done till all metadata is processed,
so the next "single step" done will be for the audio frame. After
this, we should keep to 1 input buffer -> 1 output buffer and so
avoid getting out of sync.

https://bugzilla.gnome.org/show_bug.cgi?id=650960

13 years agoflacdec: bail on reserved value
Vincent Penquerc'h [Tue, 16 Aug 2011 14:32:07 +0000 (15:32 +0100)]
flacdec: bail on reserved value

Now that we look at the right bits, we can test against the reserved
value as we do for other fields.

https://bugzilla.gnome.org/show_bug.cgi?id=650960

13 years agoflacdec: fix bit twiddling
Vincent Penquerc'h [Tue, 16 Aug 2011 14:27:43 +0000 (15:27 +0100)]
flacdec: fix bit twiddling

Right shifting a 8 bit value by 8 bits is twice too much
to get the high 4 bits.

https://bugzilla.gnome.org/show_bug.cgi?id=650960

13 years agoflacdec: warn if we see a variable block size where unsupported
Vincent Penquerc'h [Tue, 16 Aug 2011 14:22:46 +0000 (15:22 +0100)]
flacdec: warn if we see a variable block size where unsupported

https://bugzilla.gnome.org/show_bug.cgi?id=650960

13 years agospectrum: avoid crashing by resetting the correct number of channels
Vincent Penquerc'h [Tue, 16 Aug 2011 17:25:29 +0000 (18:25 +0100)]
spectrum: avoid crashing by resetting the correct number of channels

https://bugzilla.gnome.org/show_bug.cgi?id=656606

13 years agoflacparse: fix off by one in frame size check
Vincent Penquerc'h [Tue, 16 Aug 2011 12:16:22 +0000 (13:16 +0100)]
flacparse: fix off by one in frame size check

Yes, I was tracking another bug and the small test file I generated
to test with improbably just happened to trigger this, with a second
and last frame of 1615 bytes.

https://bugzilla.gnome.org/show_bug.cgi?id=656649

13 years agoid3demux: remove specs from git as well now that parsing code is in -base
Tim-Philipp Müller [Sun, 14 Aug 2011 19:46:01 +0000 (20:46 +0100)]
id3demux: remove specs from git as well now that parsing code is in -base

13 years agoid3demux: use -base provided id3 tag parsing
Mark Nauwelaerts [Thu, 14 Jul 2011 13:42:36 +0000 (15:42 +0200)]
id3demux: use -base provided id3 tag parsing

https://bugzilla.gnome.org/show_bug.cgi?id=654388

13 years agojackaudiosrc: fix error message code
Tim-Philipp Müller [Sat, 13 Aug 2011 15:51:22 +0000 (16:51 +0100)]
jackaudiosrc: fix error message code

And also post 'not found' error if jackd is not even installed.

13 years agoqtdemux: initialize bitrate variable and reset for each loop
Stefan Kost [Fri, 12 Aug 2011 14:32:58 +0000 (16:32 +0200)]
qtdemux: initialize bitrate variable and reset for each loop

Don't check eventually unset variable and don't accidentially use values from last
cycle.

13 years agortspsrc: Properly error out if SDP contains no streams
Edward Hervey [Tue, 9 Aug 2011 09:28:17 +0000 (11:28 +0200)]
rtspsrc: Properly error out if SDP contains no streams

Also fixes unitialized variable error on macosx.

13 years agoximagesrc: clear flags on buffer reuse
Vincent Penquerc'h [Tue, 9 Aug 2011 08:05:31 +0000 (09:05 +0100)]
ximagesrc: clear flags on buffer reuse

This will ensure a logically new buffer does not keep flags from
a previous use of that buffer (eg, DISCONT would be set on the first
buffer, and mistakenly kept when reused).

https://bugzilla.gnome.org/show_bug.cgi?id=653709

13 years agov4l2: take care not to change the current format where appropriate
Vincent Penquerc'h [Mon, 8 Aug 2011 09:54:26 +0000 (10:54 +0100)]
v4l2: take care not to change the current format where appropriate

Some drivers are buggy are will change the current format when
processing VIDIOC_TRY_FMT. Save and restore the current format
to ensure the format is kept unchanged.

https://bugzilla.gnome.org/show_bug.cgi?id=649067

13 years agov4l2src: Use fraction compare util function.
Sjoerd Simons [Sun, 7 Aug 2011 10:23:26 +0000 (12:23 +0200)]
v4l2src: Use fraction compare util function.

Use the fraction compare utility to compare function, not the
handcrafted one. The handcrafted one is buggy as it doesn't take into
account rounding error. For example comparing a framerate of 20/1 on a
camera configured as 30/1 fps would yield true: 1 == (1 * 20)/30 and not
re-configure the camera. Fixes #656104

13 years agomatroska: Register new debug category
Jan Schmidt [Wed, 3 Aug 2011 12:50:05 +0000 (22:50 +1000)]
matroska: Register new debug category

Register the matroskareadcommon debug category when the
plugin is loaded to avoid assertion output when debug is turned on.

13 years agoqtdemux: soften assertion check on stream size
Philippe Normand [Fri, 29 Jul 2011 11:03:55 +0000 (13:03 +0200)]
qtdemux: soften assertion check on stream size

https://bugzilla.gnome.org/show_bug.cgi?id=655570

13 years agortpjpegpay: Add support for H.264 payload in MJPEG container
Robert Krakora [Wed, 3 Aug 2011 08:09:42 +0000 (10:09 +0200)]
rtpjpegpay: Add support for H.264 payload in MJPEG container

See http://www.quickcamteam.net/uvc-h264/USB_Video_Payload_H.264_0.87.pdf

Fixes bug #655530.

13 years agojackaudiosink: Don't call g_alloca() in process_cb
Tristan Matthews [Wed, 3 Aug 2011 02:05:08 +0000 (22:05 -0400)]
jackaudiosink: Don't call g_alloca() in process_cb

g_alloca() is not RT-safe, so instead we should allocate the
memory needed in advance. Fixes #655866

13 years agodocs: fix two more Since: tags
Tim-Philipp Müller [Tue, 2 Aug 2011 22:42:58 +0000 (23:42 +0100)]
docs: fix two more Since: tags

13 years agodeinterlace: Fix Since tags for fieldanalysis related new properties
Mart Raudsepp [Sun, 31 Jul 2011 01:19:00 +0000 (04:19 +0300)]
deinterlace: Fix Since tags for fieldanalysis related new properties

commit c1b100cf9c is after 0.10.29 and 0.10.30 was a branched release.
So fix Since tags from 0.10.29 to 0.10.31 for the new properties.

13 years agopulsesink: fix variable-set-but-not-used compiler warning with older pulse versions
Tim-Philipp Müller [Fri, 29 Jul 2011 12:05:42 +0000 (13:05 +0100)]
pulsesink: fix variable-set-but-not-used compiler warning with older pulse versions

13 years agortpsession: properly init rtcp_min_interval
Mark Nauwelaerts [Fri, 29 Jul 2011 10:07:24 +0000 (12:07 +0200)]
rtpsession: properly init rtcp_min_interval

13 years agopulsesink: Add support for compressed formats
Arun Raghavan [Wed, 9 Mar 2011 05:34:36 +0000 (11:04 +0530)]
pulsesink: Add support for compressed formats

This adds support for various compressed formats (AC3, E-AC3, DTS and
MP3) payloaded in IEC 61937 format (used for transmission over S/PDIF,
HDMI and Bluetooth).

The acceptcaps() function allows bins to probe for what formats the sink
being connected to support. This only works after the element is set to
at least READY.

If the underlying sink changes and the format we are streaming is not
available, we emit a message that will allow upstream elements/bins to
block and renegotiate a new format.

13 years agopulsesink: Use the extended stream API if available
Arun Raghavan [Tue, 1 Mar 2011 10:04:46 +0000 (15:34 +0530)]
pulsesink: Use the extended stream API if available

This uses the new extended API for creating streams. This will allow us
to support compressed formats natively in pulsesink as well.

13 years agopulsesrc: Add a source-output-index property
Arun Raghavan [Thu, 28 Jul 2011 18:37:52 +0000 (00:07 +0530)]
pulsesrc: Add a source-output-index property

This exposes the source output index of the record stream that we open
so that clients can use this with the introspection if they want (to
move the stream, for example).

13 years agortpssrcdemux: keep a ref on the src pad while using it
Mark Nauwelaerts [Thu, 28 Jul 2011 12:44:57 +0000 (14:44 +0200)]
rtpssrcdemux: keep a ref on the src pad while using it

Prevent a possible race if clear_ssrc() is called between getting the pad and
doing the push.

Based on patch by <olivier.crete@collabora.com>

https://bugzilla.gnome.org/show_bug.cgi?id=650916

13 years agortpssrcdemux: Make the pads lock recursive and hold it across the signal emit
Olivier Crête [Tue, 24 May 2011 08:29:57 +0000 (11:29 +0300)]
rtpssrcdemux: Make the pads lock recursive and hold it across the signal emit

We need to keep the lock held because we don't want a push before the "new-ssrc-pad"
handler has completed. But we may want to push an event from inside that handler, hence
the recursive mutex.

https://bugzilla.gnome.org/show_bug.cgi?id=650916

13 years agortpssrcdemux: Use PADs lock
Olivier Crête [Tue, 24 May 2011 08:17:25 +0000 (11:17 +0300)]
rtpssrcdemux: Use PADs lock

https://bugzilla.gnome.org/show_bug.cgi?id=650916

13 years agortph264depay: Cope with FU-A E bit not being set
Sjoerd Simons [Wed, 27 Jul 2011 17:15:20 +0000 (18:15 +0100)]
rtph264depay: Cope with FU-A E bit not being set

Some h264 payloaders are unfortunately buggy and don't correctly set the
E bit in FU-A NAL when they have ended. Work around this by assuming
such a fragmentation unit has ended when there was no packet loss and a
new NAL is started

13 years agoac3parse: Support switching alignment on-the-fly
Arun Raghavan [Tue, 12 Apr 2011 11:31:47 +0000 (17:01 +0530)]
ac3parse: Support switching alignment on-the-fly

This allows switching of alignment for E-AC3 streams at run-time. This
is requested by downstream elements via a custom event.

https://bugzilla.gnome.org/show_bug.cgi?id=650313

13 years agoac3parse: Add support for IEC 61937 alignment
Arun Raghavan [Sat, 9 Apr 2011 06:56:56 +0000 (12:26 +0530)]
ac3parse: Add support for IEC 61937 alignment

When pushing out buffers over S/PDIF or HDMI, IEC 61937 payloading
requires each buffer to contain 6 blocks from each substream. This adds
code to collect all the frames needed to meet this requirement before
pushing out a buffer.

https://bugzilla.gnome.org/show_bug.cgi?id=650313

13 years agortpsession: Always send application requested feedback in immediate mode
Olivier Crête [Wed, 8 Jun 2011 19:57:37 +0000 (15:57 -0400)]
rtpsession: Always send application requested feedback in immediate mode

Send as many application requested feedback messages in immediate mode, even if they
have already been sent.

https://bugzilla.gnome.org/show_bug.cgi?id=654583

13 years agortpsession: Don't let the computed RTP bandwidth fall too low
Olivier Crête [Wed, 8 Jun 2011 18:48:01 +0000 (14:48 -0400)]
rtpsession: Don't let the computed RTP bandwidth fall too low

If it falls too low, the computed RTCP bandwidth will be near zero and
the RTCP thread will be stopped.

https://bugzilla.gnome.org/show_bug.cgi?id=654583

13 years agortpsession: Wait longer to timeout SSRC collision
Olivier Crête [Mon, 25 Apr 2011 20:13:38 +0000 (16:13 -0400)]
rtpsession: Wait longer to timeout SSRC collision

Using the current RTCP interval to timeout SSRC collision can lead to
collisions being timed out immediately if a BYE packet is sent because
it is sent immediately, so the interval is 0. This is not what we
want. So just set a static 10 times the default RTCP interval, it
should be enough

https://bugzilla.gnome.org/show_bug.cgi?id=648642

13 years agortspsrc: set SOURCE flag at init time
Mark Nauwelaerts [Tue, 19 Jul 2011 11:38:01 +0000 (13:38 +0200)]
rtspsrc: set SOURCE flag at init time

Fixes #654816.

13 years agortph264depay: Complete merged AU on marker bit
Olivier Crête [Mon, 18 Jul 2011 20:46:27 +0000 (16:46 -0400)]
rtph264depay: Complete merged AU on marker bit

The marker bit on a RTP packet means the AU has been completed, so push it out
immediately to reduce the latency.

https://bugzilla.gnome.org/show_bug.cgi?id=654850

13 years agortph264pay: Only set the marker bit on the last NALU of a multi-NALU access unit
Olivier Crête [Tue, 19 Jul 2011 00:27:38 +0000 (20:27 -0400)]
rtph264pay: Only set the marker bit on the last NALU of a multi-NALU access unit

An access unit could contain multiple NAL units, in that case, only the last
RTP packet of the last NALU should have its marker bit set.

https://bugzilla.gnome.org/show_bug.cgi?id=654850

13 years agomultipart: fix compiler warning
Alessandro Decina [Wed, 20 Jul 2011 06:52:58 +0000 (08:52 +0200)]
multipart: fix compiler warning

13 years agoauparse: avoid hanging on invalid short input
Mark Nauwelaerts [Tue, 19 Jul 2011 10:05:51 +0000 (12:05 +0200)]
auparse: avoid hanging on invalid short input

... as in such case there is no srcpad yet on which to forward EOS.

13 years agopulsesrc: Fix default value leaking
Thiago Santos [Mon, 18 Jul 2011 18:13:33 +0000 (15:13 -0300)]
pulsesrc: Fix default value leaking

Remember to free the default value of client name, avoiding a
leak

13 years agortph264depay: reset upon FLUSH_STOP
Mark Nauwelaerts [Mon, 18 Jul 2011 12:24:48 +0000 (14:24 +0200)]
rtph264depay: reset upon FLUSH_STOP

... which is particularly needed when merging NAL units, where not resetting
would lead to output of an older (pre-flush) AU (with unintended timestamp).

13 years agomultifilesink: do not use g_slist_free_full
Mark Nauwelaerts [Mon, 18 Jul 2011 12:30:51 +0000 (14:30 +0200)]
multifilesink: do not use g_slist_free_full

... as that is only in GLib 2.28, which is not yet required at this time.

13 years agomultifilesink: add max-files property
Alessandro Decina [Mon, 18 Jul 2011 07:38:26 +0000 (09:38 +0200)]
multifilesink: add max-files property

Add max-files property to limit the number of files saved on disk.

API: multifilesink::max-files

13 years agomultifilesink: refactor file opening and closing code
Alessandro Decina [Sun, 17 Jul 2011 21:36:55 +0000 (23:36 +0200)]
multifilesink: refactor file opening and closing code

13 years agomatroskademux: fix pixel-aspect-ratio if header has only one display variable
Alexey Fisher [Sat, 16 Jul 2011 17:38:51 +0000 (19:38 +0200)]
matroskademux: fix pixel-aspect-ratio if header has only one display variable

Current matroska demux calculates the pixel aspect ratio only if both
DisplayHeight and DisplayWidth are set, but it is legal to use only
one variable if the other is equal to PixelWidth or PixelHeight, at
least the mkclean utility is doing that. So this makse mkcleaned
files play correctly.

https://bugzilla.gnome.org/show_bug.cgi?id=654744

13 years agogoom: fix build on PPC on openbsd
Antoine Jacoutot [Sat, 16 Jul 2011 22:47:50 +0000 (23:47 +0100)]
goom: fix build on PPC on openbsd

A missing sys/param.h include results in:

/usr/include/sys/proc.h:64: error: 'MAXLOGNAME' undeclared here (not in a
function)
/usr/include/sys/proc.h:285: error: 'MAXCOMLEN' undeclared here (not in a
function)

when compiling goom on openbsd/ppc. We can just remove the two sys/ includes
here, they are not needed for anything.

https://bugzilla.gnome.org/show_bug.cgi?id=654749

13 years agortppcmApay/depay: Static clock rates on static payloads, dynamic on dynamic
Olivier Crête [Fri, 15 Jul 2011 00:10:02 +0000 (20:10 -0400)]
rtppcmApay/depay: Static clock rates on static payloads, dynamic on dynamic

Partially reverts 397dc60b

13 years agortph264pay: Implement getcaps
Olivier Crête [Fri, 4 Mar 2011 20:41:22 +0000 (15:41 -0500)]
rtph264pay: Implement getcaps

Convert profile-level-id from RTP caps into video/x-h264 style caps (with profile and level)

13 years agortspsrc: fix seeking regression
Mark Nauwelaerts [Tue, 12 Jul 2011 13:04:38 +0000 (15:04 +0200)]
rtspsrc: fix seeking regression

... introduced when shuffling around code for the async implementation
by setting state of source (and udp sources) in _play before downstream
flushing is undone.

13 years agoac3parse: fix buffer duration on blocks-per-frame change
René Stadler [Mon, 11 Jul 2011 12:23:41 +0000 (15:23 +0300)]
ac3parse: fix buffer duration on blocks-per-frame change

The gst_base_parse_set_frame_rate call was predicated on a change to
sample rate, duration or profile. However, the block count per frame can
also change between packets, which would result in incorrect buffer
durations.

13 years agomultifilesrc: Improve looping
David Schleef [Sun, 10 Jul 2011 02:23:41 +0000 (19:23 -0700)]
multifilesrc: Improve looping

Add start-index and stop-index properties.

13 years agomultifile: add loop property to multifilesrc
Jonny Lamb [Thu, 16 Jun 2011 12:57:03 +0000 (13:57 +0100)]
multifile: add loop property to multifilesrc

Fixes: #652727

Signed-off-by: Jonny Lamb <jonnylamb@jonnylamb.com>
Signed-off-by: David Schleef <ds@schleef.org>
13 years agodirectsoundsink: 16-bit audio is signed, 8-bit is unsigned.
Philip Jägenstedt [Fri, 20 Nov 2009 09:07:43 +0000 (10:07 +0100)]
directsoundsink: 16-bit audio is signed, 8-bit is unsigned.

Pretending to handle 8-bit signed causes distorted audio when
actually given such audio, which you will get if passing 8-bit
unsigned through audioconvert ! audioresample, as audioresample
only handles 8-bit signed.  Fixes #605834.

Signed-off-by: David Schleef <ds@schleef.org>
13 years agomatroskademux: handle blocks with duration=0
Alexey Fisher [Thu, 7 Jul 2011 16:27:36 +0000 (18:27 +0200)]
matroskademux: handle blocks with duration=0

Some video frames, for example alt-ref frame in VP8, will be
never displayed. This is why it has duration=0.

This patch allow to use this duration.

Bug: 654175
Signed-off-by: Alexey Fisher <bug-track@fisher-privat.net>
13 years agoqtmux: Add direct dirac mapping
David Schleef [Thu, 7 Jul 2011 00:18:05 +0000 (17:18 -0700)]
qtmux: Add direct dirac mapping

13 years agopulsesink: prevent race condition causing ref leak
René Stadler [Wed, 29 Jun 2011 17:59:26 +0000 (20:59 +0300)]
pulsesink: prevent race condition causing ref leak

Since commit 8bfd80, gst_pulseringbuffer_stop doesn't wait for the
deferred call to be run before returning. This causes a race when
READY->NULL is executed shortly after, which stops the mainloop. This
leaks the element reference which is passed as userdata for the callback
(introduced in commit 7cf996, bug #614765).

The correct fix is to wait in READY->NULL for all outstanding calls to
be fired (since libpulse doesn't provide a DestroyNotify for the
userdata). We get rid of the reference passing from 7cf996 altogether,
since finalization from the callback would anyways lead to a deadlock.

Re-fixes bug #614765.

13 years agopulsesink: small cleanup of copy-paste code
René Stadler [Mon, 4 Jul 2011 05:58:14 +0000 (08:58 +0300)]
pulsesink: small cleanup of copy-paste code

13 years agopulsesink: remove unused member variable and misleading log message
René Stadler [Wed, 29 Jun 2011 16:50:42 +0000 (19:50 +0300)]
pulsesink: remove unused member variable and misleading log message

Wim changed it in commit 8bfd80 so that pa_defer_ran is not read
anywhere.

The log message used to annotate a mainloop_wait call which is gone.

13 years agogoom: Don't answer lantency queries before negotiation
David Schleef [Mon, 4 Jul 2011 19:58:38 +0000 (12:58 -0700)]
goom: Don't answer lantency queries before negotiation

13 years agojpegdec: avoid crashing on invalid input without components
Mark Nauwelaerts [Mon, 4 Jul 2011 12:30:09 +0000 (14:30 +0200)]
jpegdec: avoid crashing on invalid input without components

13 years agoflvmux: pass along segment info to collectpads
Mark Nauwelaerts [Mon, 4 Jul 2011 09:25:28 +0000 (11:25 +0200)]
flvmux: pass along segment info to collectpads

... so it can track this and be subsequently used to determine running time etc.

13 years agoflvdemux: indicate raw format in aac caps
Mark Nauwelaerts [Mon, 4 Jul 2011 09:24:23 +0000 (11:24 +0200)]
flvdemux: indicate raw format in aac caps

13 years agopulse: Increase ranks to PRIMARY + 10
David Schleef [Mon, 4 Jul 2011 02:51:32 +0000 (19:51 -0700)]
pulse: Increase ranks to PRIMARY + 10

So that pulsesrc/pulsesink get chosen over other possible PRIMARY
src/sinks by autoaudiosink.  Presumably, if pulse is available, it
is always preferred over another src/sink.

Fixes: #647540.

13 years agomultipartmux: Add \r\n to tail of pushed buffers
David Schleef [Fri, 1 Jul 2011 01:47:48 +0000 (18:47 -0700)]
multipartmux: Add \r\n to tail of pushed buffers

Clients such as Firefox require the \r\n after the payload.

13 years agomatroskademux: avoid looping when searching for clusters
Branko Subasic [Thu, 16 Jun 2011 12:52:51 +0000 (14:52 +0200)]
matroskademux: avoid looping when searching for clusters

Fixes some bugs that results in the demuxer looping when seaching
for clusters in non-finalized files.

https://bugzilla.gnome.org/show_bug.cgi?id=652195

13 years agomatroskaparse: fix reference counting of parse->streamheader
Debarshi Ray [Fri, 10 Jun 2011 13:24:48 +0000 (18:54 +0530)]
matroskaparse: fix reference counting of parse->streamheader

https://bugzilla.gnome.org/show_bug.cgi?id=652286

Signed-off-by: David Schleef <ds@schleef.org>