platform/upstream/gstreamer.git
13 years agodiscoverer: Chain dispose() up to parent class
Arun Raghavan [Tue, 8 Feb 2011 07:12:32 +0000 (12:42 +0530)]
discoverer: Chain dispose() up to parent class

13 years agodiscoverer: Keep a ref for the async timeout callback
Arun Raghavan [Mon, 7 Feb 2011 07:34:55 +0000 (13:04 +0530)]
discoverer: Keep a ref for the async timeout callback

This makes sure we maintain a ref on the discoverer object while the
async timeout callback is alive to prevent a potential crash if the
object is freed while the callback is pending.

https://bugzilla.gnome.org/show_bug.cgi?id=641706

13 years agodiscoverer: Use g_signal_connect_object instead of g_signal_connect
Arun Raghavan [Mon, 7 Feb 2011 08:27:39 +0000 (13:57 +0530)]
discoverer: Use g_signal_connect_object instead of g_signal_connect

We want to make sure the discoverer object passed to the various
callbacks doesn't become invalid if a callback is pending and the object
is free'd in the mean time.

https://bugzilla.gnome.org/show_bug.cgi?id=641706

13 years agotypefinding: detect raw h.263
Parthasarathi Susarla [Wed, 9 Feb 2011 16:22:42 +0000 (03:22 +1100)]
typefinding: detect raw h.263

https://bugzilla.gnome.org/show_bug.cgi?id=623846

13 years agopbutils: encoding-target: fix error checking in target file loading
Teemu Katajisto [Mon, 21 Feb 2011 13:58:16 +0000 (15:58 +0200)]
pbutils: encoding-target: fix error checking in target file loading

https://bugzilla.gnome.org/show_bug.cgi?id=642949

13 years agotests: fix videoscale test by ignoring newly-added 64-bit formats
Tim-Philipp Müller [Mon, 21 Feb 2011 17:55:04 +0000 (17:55 +0000)]
tests: fix videoscale test by ignoring newly-added 64-bit formats

They probably fail because ffmpegcolorspace can't handle those formats.

13 years agosdp: Fix copy/paste error in inrospection part of Makefile
Benjamin Otte [Mon, 21 Feb 2011 17:01:04 +0000 (18:01 +0100)]
sdp: Fix copy/paste error in inrospection part of Makefile

13 years agotag: Fix copy/paste error in inrospection part of Makefile
Benjamin Otte [Mon, 21 Feb 2011 17:00:36 +0000 (18:00 +0100)]
tag: Fix copy/paste error in inrospection part of Makefile

13 years agortsp: Fix copy/paste error in inrospection part of Makefile
Benjamin Otte [Mon, 21 Feb 2011 17:00:02 +0000 (18:00 +0100)]
rtsp: Fix copy/paste error in inrospection part of Makefile

13 years agoaudiorate: add skip-to-first property
Mark Nauwelaerts [Mon, 21 Feb 2011 11:40:36 +0000 (12:40 +0100)]
audiorate: add skip-to-first property

API: GstAudioRate::skip-to-first

13 years agovideorate: fix skip-to-first ts setup
Mark Nauwelaerts [Mon, 21 Feb 2011 11:27:17 +0000 (12:27 +0100)]
videorate: fix skip-to-first ts setup

... such as avoiding arithmetic mixing counts and ts, although latter
would typically be 0 so far.

13 years agoRevert "oggmux,adder: Check if collectpads has been freed"
Edward Hervey [Mon, 21 Feb 2011 11:04:09 +0000 (12:04 +0100)]
Revert "oggmux,adder: Check if collectpads has been freed"

This reverts commit 6d150873e8b4c23d694b0351570de323b1576d76.

Depends on a core commit that was reverted.

13 years agooggmux,adder: Check if collectpads has been freed
David Schleef [Mon, 21 Feb 2011 07:49:54 +0000 (23:49 -0800)]
oggmux,adder: Check if collectpads has been freed

Core now calls release_pad in finalize, which is usually after
the collectpads has been unreffed.

13 years agovideoscale: Add 16-bit-channel support
David Schleef [Sun, 20 Feb 2011 02:50:37 +0000 (18:50 -0800)]
videoscale: Add 16-bit-channel support

13 years agovideotestsrc: Add 16-bit-per-channel formats
David Schleef [Sun, 20 Feb 2011 00:41:43 +0000 (16:41 -0800)]
videotestsrc: Add 16-bit-per-channel formats

13 years agovideo: Add ARGB64 and AYUV64
David Schleef [Sat, 19 Feb 2011 20:03:17 +0000 (12:03 -0800)]
video: Add ARGB64 and AYUV64

16-bit per channel formats.

13 years agovideo: Add gst_video_format_get_component_depth()
David Schleef [Sat, 19 Feb 2011 00:26:59 +0000 (16:26 -0800)]
video: Add gst_video_format_get_component_depth()

13 years agoaudiotestsrc: each element gets its own instance of GRand, if needed
Leo Singer [Fri, 18 Feb 2011 21:27:23 +0000 (13:27 -0800)]
audiotestsrc: each element gets its own instance of GRand, if needed

As a result, pipelines that contain multiple instances of audiotestsrc
with the 'wave' property set to 'white-noise', 'pink-noise', or
'gaussian-noise' will run much faster, since they won't be competing
for access to the global, lock-protected instance of GRand.

Fixes bug #642720.

13 years agoplaybin2: If a sink claims to support ANY caps assume that it only supports the usual...
Sebastian Dröge [Fri, 18 Feb 2011 16:26:53 +0000 (17:26 +0100)]
playbin2: If a sink claims to support ANY caps assume that it only supports the usual raw formats

This should be changed again in 0.11, if a sink really claims to support ANY
caps it should support everything or provide correct caps.

13 years agoencodebin: Add a audioconverter after the audio resampler.
Edward Hervey [Thu, 17 Feb 2011 17:11:10 +0000 (18:11 +0100)]
encodebin: Add a audioconverter after the audio resampler.

This allows handling non-native-endianness conversion properly.

13 years agoplaybin2: Use gst_pad_accept_caps() instead of intersecting with the getcaps caps
Sebastian Dröge [Fri, 18 Feb 2011 13:04:38 +0000 (14:04 +0100)]
playbin2: Use gst_pad_accept_caps() instead of intersecting with the getcaps caps

This might be faster and more accurate in some cases to detect if a
sink supports a format and autoplugging can be stopped.

13 years agouridecodebin: Add default handler for autoplug-select
Sebastian Dröge [Fri, 18 Feb 2011 11:06:30 +0000 (12:06 +0100)]
uridecodebin: Add default handler for autoplug-select

uridecodebin proxies this signal and only the first signal handler
will ever be called from decodebin2, which is uridecodebin's proxy
signal handler.

13 years agouridecodebin: Return NULL from the default autoplug-sort handler
Sebastian Dröge [Fri, 18 Feb 2011 11:02:07 +0000 (12:02 +0100)]
uridecodebin: Return NULL from the default autoplug-sort handler

...instead of copying the array. Returning NULL will result
in the original factories array to be used and prevents a useless
array copy in most use cases.

13 years agodecodebin2: Return NULL from the default autoplug-sort handler
Sebastian Dröge [Fri, 18 Feb 2011 11:01:05 +0000 (12:01 +0100)]
decodebin2: Return NULL from the default autoplug-sort handler

...instead of copying the array. Returning NULL will result
in the original factories array to be used and prevents a useless
array copy in most use cases.

13 years agouridecodebin: Update autoplug-* signal docs from decodebin2
Sebastian Dröge [Fri, 18 Feb 2011 11:00:34 +0000 (12:00 +0100)]
uridecodebin: Update autoplug-* signal docs from decodebin2

uridecodebin proxies these signals.

13 years agodecodebin2: Update documentation of the autoplug-* signals
Sebastian Dröge [Fri, 18 Feb 2011 10:58:44 +0000 (11:58 +0100)]
decodebin2: Update documentation of the autoplug-* signals

Add notes about the behaviour if multiple signal handlers are connected.
For most autoplug-* signals only the first signal handler will ever
be invoked.

Also add to the autoplug-sort docs that the signal handler can return NULL
to specify that the order should change and other handlers get the chance
to sort the array.

13 years agodecodebin2: Keep the original factory list if the sort signal handlers returned NULL
Sebastian Dröge [Fri, 18 Feb 2011 10:57:12 +0000 (11:57 +0100)]
decodebin2: Keep the original factory list if the sort signal handlers returned NULL

13 years agouridecodebin: expose "autoplug-sort" signal
tskd2@yahoo.co.jp [Wed, 16 Feb 2011 11:14:25 +0000 (20:14 +0900)]
uridecodebin: expose "autoplug-sort" signal

It is a proxy of the decodebin2's one, and was missing
in the previous code.

See bug #642433.

13 years agoplaybin2: Use a recursive mutex for the playbin lock
Sebastian Dröge [Fri, 18 Feb 2011 09:57:40 +0000 (10:57 +0100)]
playbin2: Use a recursive mutex for the playbin lock

This lock is taken when activating a group, which could result in
calling the autoplug-continue callback, which also needs this lock
to access the sinks.

See bug #642174.

13 years agodecodebin2: Disconnect signal handlers when removing a failed element
Sebastian Dröge [Fri, 18 Feb 2011 08:36:34 +0000 (09:36 +0100)]
decodebin2: Disconnect signal handlers when removing a failed element

This prevents crashes later if one of the signals is emitted after the
element was removed from decodebin2 already, which can happen in discoverer.

13 years agotypefind: Fix mpeg TS detection
David Schleef [Wed, 16 Feb 2011 03:23:48 +0000 (19:23 -0800)]
typefind: Fix mpeg TS detection

13 years agotheoraenc: move debug category init earlier
David Schleef [Sat, 5 Feb 2011 01:36:40 +0000 (17:36 -0800)]
theoraenc: move debug category init earlier

13 years agooggparse: better detection of delta unit flag
David Schleef [Fri, 4 Feb 2011 06:41:23 +0000 (22:41 -0800)]
oggparse: better detection of delta unit flag

13 years agotheoraenc: Set speed level while running
David Schleef [Sun, 16 Jan 2011 02:21:28 +0000 (18:21 -0800)]
theoraenc: Set speed level while running

13 years agoSet the theoraenc speed-level property from libtheora's defaults.
Ralph Giles [Thu, 13 Jan 2011 23:12:53 +0000 (15:12 -0800)]
Set the theoraenc speed-level property from libtheora's defaults.

The speed-level property, which allows callers to trade of encoding
quality for speed in the libtheora api, has a version-dependent
maximum and default values. Instead of hardcoding the acceptable
range for the theoraenc element's presentation of this setting,
we query the library directly at class initialization time and
set the maximum and default values from that. If the query fails,
we fall back to the previous default setting.

To keep the values reported by gst-inspect (which I'm told use
the spec values from the class) with those available on an\
instantiated element, we remove to setting of enc->speed_level
from the initializer and instead pass G_PARAM_CONSTRUCT to
the property spec flags, asking g_object to set this property
when theoraenc objects are constructed.

NB in theory the maximum speed-level could depend on the actual
video caps. If later versions of libtheoraenc do this, a second
call will need to be made from theora_enc_reset to update the
property, since this function is mostly useful for realtime
adjustment of performance while the pipeline is running.

13 years agodiscoverer: don't leak parent tags
Stefan Kost [Wed, 16 Feb 2011 09:57:31 +0000 (11:57 +0200)]
discoverer: don't leak parent tags

13 years agodiscoverer: improve logging (and reindent)
Stefan Kost [Wed, 16 Feb 2011 09:56:16 +0000 (11:56 +0200)]
discoverer: improve logging (and reindent)

Add more logging for the tag merging and use the _OBJECT flavour more.

13 years agoplaybin2: Optimize autoplug-continue handler a bit
Sebastian Dröge [Tue, 15 Feb 2011 16:46:22 +0000 (17:46 +0100)]
playbin2: Optimize autoplug-continue handler a bit

Don't build merge the caps of all sinks but check them one-by-one
until one supports the caps. Also get reffed caps from the sinkpads
instead of a writable copy and add debug output if a sink claims to
support ANY caps.

13 years agoplaybin2: Fix handling of non-raw custom sinks
Akihiro Tsukada [Tue, 15 Feb 2011 16:24:28 +0000 (17:24 +0100)]
playbin2: Fix handling of non-raw custom sinks

When autoplugging elements in decodebin2, check if
the caps are supported by one of the sink before
continuing autoplugging.

Fixes bug #642174.

13 years agodecodebin2: Don't leak elements that fail to go to PAUSED after being autoplugged
Sebastian Dröge [Tue, 15 Feb 2011 16:01:13 +0000 (17:01 +0100)]
decodebin2: Don't leak elements that fail to go to PAUSED after being autoplugged

Fixes bug #642381.

13 years agotheoraenc: Don't reset the video quality setting the bitrate
Sjoerd Simons [Sun, 13 Feb 2011 14:42:14 +0000 (14:42 +0000)]
theoraenc: Don't reset the video quality setting the bitrate

libtheora has two encoding modes, CBR, where it tries to hit a target
bitrate and VBR where it tries to achieve a target quality.

Internally if the target bitrate is set to anything other then 0 the
encoding-mode is CBR.

This means that the gstreamer element can leave the video_quality
setting alone as long as the user is tweaking the bitrate. Which has the
nice side-effect that if the user explicitely sets the bitrate to 0
(which is actually the default), the quality value doesn't get reset and
one ends up encoding VBR at quality-level 0...

13 years agogdppay: ensure buffer's metadata is writable before setting caps
Andoni Morales Alastruey [Wed, 9 Feb 2011 11:45:23 +0000 (12:45 +0100)]
gdppay: ensure buffer's metadata is writable before setting caps

13 years agoAutomatic update of common submodule
Stefan Kost [Mon, 14 Feb 2011 10:52:59 +0000 (12:52 +0200)]
Automatic update of common submodule

From f94d739 to 1de7f6a

13 years agodoap: update mailing list location
Tim-Philipp Müller [Thu, 10 Feb 2011 23:44:43 +0000 (23:44 +0000)]
doap: update mailing list location

13 years agodiscoverer: Use nominal bitrate if bitrate tag is unavailable
Arun Raghavan [Tue, 8 Feb 2011 18:28:56 +0000 (23:58 +0530)]
discoverer: Use nominal bitrate if bitrate tag is unavailable

If the bitrate tag is unavailable, this falls back to the nominal
bitrate tag instead, if that is present.

https://bugzilla.gnome.org/show_bug.cgi?id=641860

13 years agodecodebin2: caps can be NULL
Stefan Kost [Tue, 8 Feb 2011 10:31:34 +0000 (12:31 +0200)]
decodebin2: caps can be NULL

Don't use and unref NULL caps.

13 years agotagdemux: also push cached events downstream when operating in pull mode
Mark Nauwelaerts [Wed, 2 Feb 2011 15:49:04 +0000 (16:49 +0100)]
tagdemux: also push cached events downstream when operating in pull mode

Otherwise, having 2 tagdemux in a row followed by an element operating in
pull mode will make the second tagdemux implictly eat the first tagdemux'
tag event(s).

Fixes (part of) #641047.

13 years agooggmux: ensure serialnos are unique
Vincent Penquerc'h [Fri, 21 Jan 2011 18:10:29 +0000 (18:10 +0000)]
oggmux: ensure serialnos are unique

We do that by checking a newly generated one is not already used in
an existing stream, and doing it again if it is.

https://bugzilla.gnome.org/show_bug.cgi?id=640211

13 years agooggmux: free stream map caps when done
Tim-Philipp Müller [Wed, 2 Feb 2011 17:30:15 +0000 (17:30 +0000)]
oggmux: free stream map caps when done

13 years agooggmux: keep IN_CAPS flag check for header buffers as fallback
Tim-Philipp Müller [Wed, 2 Feb 2011 17:23:43 +0000 (17:23 +0000)]
oggmux: keep IN_CAPS flag check for header buffers as fallback

In case the ogg mapper doesn't handle all the accepted input formats
(although it really should). Saves us error handling for that case
though. Also log caps properly.

https://bugzilla.gnome.org/show_bug.cgi?id=629196

13 years agooggmux: use oggstream for less brittleness in recognizing headers
Vincent Penquerc'h [Fri, 21 Jan 2011 16:05:46 +0000 (16:05 +0000)]
oggmux: use oggstream for less brittleness in recognizing headers

Using the IN_CAPS flag for this is brittle, and will fail if either
vorbisparse or vorbistag (which is itself based on vorbisparse) is
inserted between oggdemux and oggmux. Possibly other elements too
(eg, theoraparse, etc).
Using oggstream ensures we Get It Right More Often Than Not.

https://bugzilla.gnome.org/show_bug.cgi?id=629196

13 years agouridecodebin: fix copy-and-paste typo in property docs
Mark Nauwelaerts [Wed, 2 Feb 2011 14:33:36 +0000 (15:33 +0100)]
uridecodebin: fix copy-and-paste typo in property docs

13 years agooggmux: do not skip a pageno at start
Vincent Penquerc'h [Fri, 21 Jan 2011 10:56:00 +0000 (10:56 +0000)]
oggmux: do not skip a pageno at start

Discontinuities are automatically signalled by oggdemux at the start
of a new stream. When oggmux is yet to output actual data pages,
do not signal these discontinuities in the ogg stream.

This patch may miss some actual discontinuities at the very start of
a stream, but avoids the spurious missing pages when encoding happens
normally.

A better fix might involve finding a way to distinguish between actual
data discontinuities and discontinuities merely marking the start of
a new stream.

Fixes an issue with ogg page numbering (would skip a number for no
reason, which then looks like a packet was lost somewhere) when
re-muxing an ogg stream, e.g. when re-tagging in rhythmbox.

https://bugzilla.gnome.org/show_bug.cgi?id=629196

13 years agotheoraenc: clean up property descriptions
Tim-Philipp Müller [Tue, 1 Feb 2011 15:57:14 +0000 (15:57 +0000)]
theoraenc: clean up property descriptions

Remove "This property requires libtheora version >= 1.1" qualifiers
from property descriptions. They aren't needed any longer now that
we require libtheora >= 1.1.

13 years agoid3tag: map the ID3v2 TENC frame to GST_TAG_ENCODED_BY
Sreerenj Balachandran [Thu, 19 Aug 2010 19:31:07 +0000 (22:31 +0300)]
id3tag: map the ID3v2 TENC frame to GST_TAG_ENCODED_BY

https://bugzilla.gnome.org/show_bug.cgi?id=627268

13 years agotcp: use socklen_t where appropriate rather than specific type
Mark Nauwelaerts [Sat, 29 Jan 2011 19:43:08 +0000 (20:43 +0100)]
tcp: use socklen_t where appropriate rather than specific type

In particular, fixes Cygwin build where socklen_t is defined as int
in line with native win32 api definition.

13 years agoxmptag: cast argument to isdigit to int
Mark Nauwelaerts [Sat, 29 Jan 2011 18:40:23 +0000 (19:40 +0100)]
xmptag: cast argument to isdigit to int

... as that is the specification and fixes compilation on Cygwin:

gstxmptaag.c: In function 'read_one_tag':
gstxmptag.c:1015: error: array subscript has type 'char'

13 years agoappsink: add buffer fallback in case the application doesn't handle buffer lists
Tim-Philipp Müller [Mon, 31 Jan 2011 18:06:18 +0000 (18:06 +0000)]
appsink: add buffer fallback in case the application doesn't handle buffer lists

We shouldn't assume the application handles buffer lists, for
ease-of-use reasons and for backwards compatibility reasons.

13 years agoappsink: send new-buffer-list signal
Cai Yuanqing [Wed, 26 Jan 2011 02:32:32 +0000 (10:32 +0800)]
appsink: send new-buffer-list signal

Send new-buffer-list signal when emit-signals is TRUE

https://bugzilla.gnome.org/show_bug.cgi?id=640607

13 years agouridecodebin: also add https to buffer protocols
Wim Taymans [Thu, 20 Jan 2011 15:25:42 +0000 (16:25 +0100)]
uridecodebin: also add https to buffer protocols

HTTPS also needs buffering.

13 years agobaseaudiosink: trivial cleanups
Felipe Contreras [Sun, 30 Jan 2011 13:40:53 +0000 (15:40 +0200)]
baseaudiosink: trivial cleanups

It seems these stuff was neglected from commmit d8942e2.

Signed-off-by: Felipe Contreras <felipe.contreras@nokia.com>
13 years agowin32: fix DEFAULT_AUDIOSINK, should be direct*sound*sink
Tim-Philipp Müller [Thu, 27 Jan 2011 15:26:25 +0000 (15:26 +0000)]
win32: fix DEFAULT_AUDIOSINK, should be direct*sound*sink

https://bugzilla.gnome.org/show_bug.cgi?id=640705

13 years agotypefinding: register H264 typefinder with H264 caps
Philippe Normand [Thu, 27 Jan 2011 11:32:35 +0000 (12:32 +0100)]
typefinding: register H264 typefinder with H264 caps

https://bugzilla.gnome.org/show_bug.cgi?id=640709

13 years agostreamsplitter: release pending events refs
Thiago Santos [Wed, 26 Jan 2011 15:16:58 +0000 (12:16 -0300)]
streamsplitter: release pending events refs

Unref pending events when disposing the streamsplitter.

Also refactor a little to replace a for with a g_list_foreach

13 years agotests: don't run encodebin test if vorbis or theora plugins aren't available
Tim-Philipp Müller [Wed, 26 Jan 2011 15:42:48 +0000 (15:42 +0000)]
tests: don't run encodebin test if vorbis or theora plugins aren't available

13 years agopbutils: add description for degas images
Tim-Philipp Müller [Wed, 26 Jan 2011 09:07:26 +0000 (09:07 +0000)]
pbutils: add description for degas images

13 years agotypefinding: use image/x-icon as media type for ICON files
Tim-Philipp Müller [Wed, 26 Jan 2011 09:06:10 +0000 (09:06 +0000)]
typefinding: use image/x-icon as media type for ICON files

That's what we've been using so far (e.g. gdkpixbufdec).

13 years agosnapshot: use a keyframe seek
Stefan Kost [Tue, 18 Jan 2011 08:20:29 +0000 (10:20 +0200)]
snapshot: use a keyframe seek

One would usualy get good quality snapshots quickly. The exact seek position
does not really matter.

13 years agosnapshot: add a newline to the usage and error output
Stefan Kost [Mon, 17 Jan 2011 21:13:29 +0000 (23:13 +0200)]
snapshot: add a newline to the usage and error output

13 years agodecodebin2: add comment and whitespace trimming
Stefan Kost [Tue, 25 Jan 2011 16:03:23 +0000 (18:03 +0200)]
decodebin2: add comment and whitespace trimming

13 years agopbutils: add description for x-annodex
Stefan Kost [Wed, 12 Jan 2011 12:03:12 +0000 (14:03 +0200)]
pbutils: add description for x-annodex

13 years agotypefind: add typefinder for DEGAS images
Vincent Penquerc'h [Tue, 25 Jan 2011 13:39:25 +0000 (13:39 +0000)]
typefind: add typefinder for DEGAS images

This fixes at least one DEGAS image from being misdetected as DTS audio.

https://bugzilla.gnome.org/show_bug.cgi?id=625129

13 years agooggmux: cleanup
Vincent Penquerc'h [Fri, 21 Jan 2011 14:56:28 +0000 (14:56 +0000)]
oggmux: cleanup

Remove a pointless string concatentation, and fix an off-by-one in
packetno in a log.

https://bugzilla.gnome.org/show_bug.cgi?id=640189

13 years agotypefind: add detection for windows icon files to get them out of the way
Vincent Penquerc'h [Mon, 24 Jan 2011 11:45:21 +0000 (11:45 +0000)]
typefind: add detection for windows icon files to get them out of the way

Some of them can otherwise be misdetected for MPEG audio.

https://bugzilla.gnome.org/show_bug.cgi?id=620364

13 years agooggdemux: Remove dead code
Sreerenj Balachandran [Mon, 17 Jan 2011 13:11:15 +0000 (15:11 +0200)]
oggdemux: Remove dead code

13 years agooggparse: Make gst_ogg_parse_submit_buffer() safe
Yang Xichuan [Tue, 11 Jan 2011 07:10:42 +0000 (15:10 +0800)]
oggparse: Make gst_ogg_parse_submit_buffer() safe

By not passing zero-sized buffers to ogg_sync_buffer()
and checking the return values of libogg functions.

Fixes bug #639136.

13 years agotextoverlay: Add support for vertical center alignment
Lane Brooks [Tue, 11 Jan 2011 17:18:34 +0000 (18:18 +0100)]
textoverlay: Add support for vertical center alignment

Fixes bug #639159.

13 years agotag: xmp: Move static variable to local function
Thiago Santos [Mon, 24 Jan 2011 18:21:10 +0000 (15:21 -0300)]
tag: xmp: Move static variable to local function

Variable was being written to and could cause crashes
if multiple elements were parsing xmp at the same time.

Moving it to local scope solves the problem.

13 years agoriff: Add support for video/x-camstudio
Edward Hervey [Mon, 24 Jan 2011 17:27:30 +0000 (18:27 +0100)]
riff: Add support for video/x-camstudio

13 years agoBack to development
Tim-Philipp Müller [Mon, 24 Jan 2011 00:00:27 +0000 (00:00 +0000)]
Back to development

13 years agoRelease 0.10.32
Tim-Philipp Müller [Fri, 21 Jan 2011 10:50:06 +0000 (10:50 +0000)]
Release 0.10.32

13 years ago0.10.31.4 pre-releases
Tim-Philipp Müller [Tue, 18 Jan 2011 10:45:01 +0000 (10:45 +0000)]
0.10.31.4 pre-releases

13 years agodocs: update docs
Tim-Philipp Müller [Tue, 18 Jan 2011 10:44:01 +0000 (10:44 +0000)]
docs: update docs

13 years agoencoding-target: change keyfile header to 'GStreamer Encoding Target'
Tim-Philipp Müller [Tue, 18 Jan 2011 10:40:29 +0000 (10:40 +0000)]
encoding-target: change keyfile header to 'GStreamer Encoding Target'

which is more in line with other files such as .desktop files.

13 years agopbutils: don't assume LC_MESSAGES is always defined, also check for ENABLE_NLS
Tim-Philipp Müller [Tue, 18 Jan 2011 01:06:50 +0000 (01:06 +0000)]
pbutils: don't assume LC_MESSAGES is always defined, also check for ENABLE_NLS

Should fix build with mingw32 build bot again.

13 years agoapp: export gst_app_stream_type_get_type()
Tim-Philipp Müller [Tue, 18 Jan 2011 00:09:37 +0000 (00:09 +0000)]
app: export gst_app_stream_type_get_type()

API: gst_app_stream_type_get_type()
API: GST_TYPE_APP_STREAM_TYPE

https://bugzilla.gnome.org/show_bug.cgi?id=639747

13 years agoapp: make GstAppBuffer get_type() function thread-safe
Tim-Philipp Müller [Mon, 17 Jan 2011 23:59:48 +0000 (23:59 +0000)]
app: make GstAppBuffer get_type() function thread-safe

13 years agodiscoverer: Drop new stream tags once preroll is done
Arun Raghavan [Mon, 17 Jan 2011 19:39:53 +0000 (01:09 +0530)]
discoverer: Drop new stream tags once preroll is done

This makes sure we do not touch the stream taglist once the pipeline has
been prerolled. Adding of stream tags happens in the pad event probe
which runs in a different thread from discoverer stream processing, so
modifying the tag list while discoverer might be processing it can
sometimes cause a crash.

https://bugzilla.gnome.org/show_bug.cgi?id=639778

13 years agodiscoverer: Validate timeouts before processing them
Arun Raghavan [Mon, 17 Jan 2011 10:00:08 +0000 (15:30 +0530)]
discoverer: Validate timeouts before processing them

This avoids a race where the timeout callback is scheduled to run but we
get sufficient information to finish discovery before actually getting
around to executing the callback. See the documentation of
g_source_is_destroyed() for more details.

https://bugzilla.gnome.org/show_bug.cgi?id=639730

13 years agodiscoverer: Make sure we call _stop() before being freed
Arun Raghavan [Mon, 17 Jan 2011 18:38:32 +0000 (00:08 +0530)]
discoverer: Make sure we call _stop() before being freed

This ensures that everything is properly cleaned up before the
GstDiscoverer object is freed. Specifically, it makes sure that we've
removed the async timeout callback before freeing the object to avoid a
potential crash later on.

https://bugzilla.gnome.org/show_bug.cgi?id=639755

13 years agogdppay: make newsegment buffer metadata writable
David Schleef [Sun, 16 Jan 2011 22:55:46 +0000 (14:55 -0800)]
gdppay: make newsegment buffer metadata writable

13 years agopbutils: save localised strings properly when writing encoding targets to a file
Tim-Philipp Müller [Sun, 16 Jan 2011 16:46:22 +0000 (16:46 +0000)]
pbutils: save localised strings properly when writing encoding targets to a file

Use LC_MESSAGES rather than LC_ALL. Save/load description as untranslated string
when using an English language locale. Strip locale information to the language,
so we don't save keys like description[fr_FR.UTF-8]=...

https://bugzilla.gnome.org/show_bug.cgi?id=638860

13 years agotypefinding: set framed=false on DTS caps
Tim-Philipp Müller [Thu, 13 Jan 2011 13:59:41 +0000 (13:59 +0000)]
typefinding: set framed=false on DTS caps

13 years agodocs: add some more Since: markers for new encoding-profile API
Tim-Philipp Müller [Wed, 12 Jan 2011 17:51:43 +0000 (17:51 +0000)]
docs: add some more Since: markers for new encoding-profile API

13 years agoconfigure: require gobject-introspection >= 0.9.12
Tim-Philipp Müller [Wed, 12 Jan 2011 15:51:52 +0000 (15:51 +0000)]
configure: require gobject-introspection >= 0.9.12

Earlier versions don't honour the -L/--library-path option,
which we need. See commit 4d0ccdad in gobject-introspection git.

Should "fix" build on lucid/maverick build bots.

13 years ago0.10.31.3 pre-release
Tim-Philipp Müller [Tue, 11 Jan 2011 19:19:50 +0000 (19:19 +0000)]
0.10.31.3 pre-release

13 years agopo: update translations
Tim-Philipp Müller [Tue, 11 Jan 2011 18:59:39 +0000 (18:59 +0000)]
po: update translations

13 years agoexamples: allow building with newer GTK+
Bastien Nocera [Tue, 11 Jan 2011 14:41:53 +0000 (14:41 +0000)]
examples: allow building with newer GTK+

GtkFunction is gone, and there's no update policies for
GtkRanges any more (but the default was continuous anyway,
so no need to set it to that mode explicitly).

https://bugzilla.gnome.org/show_bug.cgi?id=639215

13 years agogobject-introspection: pass --library-path as well to make it find the right libgstreamer
Tim-Philipp Müller [Tue, 11 Jan 2011 14:59:38 +0000 (14:59 +0000)]
gobject-introspection: pass --library-path as well to make it find the right libgstreamer

Makes things work again properly in uninstalled setups (and
presumably in installed setups where GStreamer is installed
into a non-standard prefix). Requires fixes from core git.

https://bugzilla.gnome.org/show_bug.cgi?id=639039

13 years agogobject-introspection: fix issue when gold linker is used
Byeong-ryeol Kim [Tue, 11 Jan 2011 14:52:51 +0000 (14:52 +0000)]
gobject-introspection: fix issue when gold linker is used

Need to pass libgstreamer-0.10 explicitly to linker, since we're
calling gst_init(), which in turn is needed because the encoding
target get_type() function calls gst_value_register().

https://bugzilla.gnome.org/show_bug.cgi?id=639039