platform/upstream/gstreamer.git
14 years agogst: Don't use GST_DEBUG_FUNCPTR for GObject vfuncs
Sebastian Dröge [Sun, 6 Jun 2010 15:52:40 +0000 (17:52 +0200)]
gst: Don't use GST_DEBUG_FUNCPTR for GObject vfuncs

14 years agomatroskademux: refactor delta unit handling
Philip Jägenstedt [Sun, 6 Jun 2010 13:12:16 +0000 (15:12 +0200)]
matroskademux: refactor delta unit handling

This allows us to skip delta units earlier and is a bit clearer in my
opinion. It also makes only video buffers ever be delta units, not
just for SimpleBlock as before.

14 years agoflvdemux: Clear adapter on discontinuities
Sebastian Dröge [Sun, 6 Jun 2010 13:17:00 +0000 (15:17 +0200)]
flvdemux: Clear adapter on discontinuities

14 years agomatroskademux: Ignore keyframe flag for non-video streams
Philip Jägenstedt [Sun, 6 Jun 2010 12:03:53 +0000 (14:03 +0200)]
matroskademux: Ignore keyframe flag for non-video streams

When the keyframe bit of SimpleBlock Flags wasn't set, the buffer was being
marked with GST_BUFFER_FLAG_DELTA_UNIT, causing all buffers to be skipped
after a seek. This may be a problem with the Sorenson Squish encoder, but
arguably the keyframe bit should only be applied to video.

Fixes bug #620358.

14 years agoflvdemux: First try upstream when handling seek events/queries
Sebastian Dröge [Sun, 6 Jun 2010 12:56:52 +0000 (14:56 +0200)]
flvdemux: First try upstream when handling seek events/queries

14 years agogstrtpceltpay: don't always fixate sink caps to 1 channel
Tristan Matthews [Fri, 4 Jun 2010 18:54:59 +0000 (14:54 -0400)]
gstrtpceltpay: don't always fixate sink caps to 1 channel

The getcaps function should not fixate the channels field until we
get the encoding-params field from our srcpad's caps. Fixes #620591

14 years agortsp: try all ranges from the sdp
Wim Taymans [Fri, 4 Jun 2010 11:57:28 +0000 (13:57 +0200)]
rtsp: try all ranges from the sdp

Try all ranges in the SDP before giving up.

14 years agortspsrc: make parse_range return result
Wim Taymans [Fri, 4 Jun 2010 11:56:07 +0000 (13:56 +0200)]
rtspsrc: make parse_range return result

Make the parse_range function return if the parsing succeeded or failed.

14 years agovideomixer: if we're not linked downstream, we can do any format
Edward Hervey [Fri, 4 Jun 2010 09:44:09 +0000 (11:44 +0200)]
videomixer: if we're not linked downstream, we can do any format

Stupid me, assuming _get_allowed_caps() would actually return the
pad templates if there was no peer.

14 years agoKeep announcing the delivery-method in the capabilities
Sjoerd Simons [Mon, 31 May 2010 15:26:19 +0000 (16:26 +0100)]
Keep announcing the delivery-method in the capabilities

Even though we don't use delivery-method in our payloader, older versions of
the theora payloader in gstreamer required it. As such we need to keep this
around in the caps for backwards-compatibility.

This reverts part of 49463a37cbaa952e1401291f0a2623de6cab3880

Fixes #618940

14 years agooss4: add some comments for translators to clarify meaning of "Low"
Tim-Philipp Müller [Thu, 3 Jun 2010 16:52:11 +0000 (17:52 +0100)]
oss4: add some comments for translators to clarify meaning of "Low"

"Low" etc. are quality settings here (e.g. for the internal resampler).
Some day when we use GLib's i18n functions we might want to use
NC_() and g_dpgettext2() here instead of the comments.

Fixes #555967.

14 years agomp4gdepay: calculate the frame duration correctly
Wim Taymans [Thu, 3 Jun 2010 17:23:01 +0000 (19:23 +0200)]
mp4gdepay: calculate the frame duration correctly

When we calculate the frame duration, we need to use the amount of
frames in the _previous_ packet, not the current packet. The frame duration is
needed to correctly de-interleave interleaved streams. This fixes the case where
there are a variable number of frames in a packet.

Fixes #620494

14 years agovideomixer: Don't return caps in get_caps() that will be rejected
Edward Hervey [Thu, 3 Jun 2010 16:58:42 +0000 (18:58 +0200)]
videomixer: Don't return caps in get_caps() that will be rejected

This commit basically puts _get_caps() in sync with accept_caps().

If we don't have a master pad OR the master pad caps aren't negotiated
then we just return the downstream allowed caps.

If we have a master pad with negotiated caps, we return those caps
with a free range of width/height/framerate

14 years agoRevert "pulsesink: Add comments to remove the provide-clock message posting once...
Sebastian Dröge [Thu, 3 Jun 2010 11:45:32 +0000 (13:45 +0200)]
Revert "pulsesink: Add comments to remove the provide-clock message posting once we depend on base 0.10.30"

This reverts commit 8f3708f38aa3839a6a625ca7d1c166101c9fbb7f.

The baseaudiosink commit was reverted

14 years agopulsesink: Add comments to remove the provide-clock message posting once we depend...
Sebastian Dröge [Thu, 3 Jun 2010 08:27:25 +0000 (10:27 +0200)]
pulsesink: Add comments to remove the provide-clock message posting once we depend on base 0.10.30

baseaudiosink does all this for us now.

14 years agoflvdemux: delayed seek handling also deserves TRUE event response
Mark Nauwelaerts [Wed, 2 Jun 2010 14:36:11 +0000 (16:36 +0200)]
flvdemux: delayed seek handling also deserves TRUE event response

14 years agortpjitterbuffer: fix compiler warning
Thijs Vermeir [Wed, 2 Jun 2010 13:30:47 +0000 (15:30 +0200)]
rtpjitterbuffer: fix compiler warning

unused variable ‘estimated’

14 years agocommon: revert the change i did in my previous commit
Alessandro Decina [Wed, 2 Jun 2010 13:04:00 +0000 (15:04 +0200)]
common: revert the change i did in my previous commit

14 years agortpjitterbuffer: stop buffering and emit EOS at the end of a stream
Alessandro Decina [Wed, 2 Jun 2010 11:39:10 +0000 (13:39 +0200)]
rtpjitterbuffer: stop buffering and emit EOS at the end of a stream

When using RTP_JITTER_BUFFER_MODE_BUFFER, make sure that the ringbuffer doesn't
get stuck buffering forever when there isn't enough data left to fill the
buffer.

14 years agodebugutils: Don't consume preroll buffer twice
Benjamin Otte [Tue, 1 Jun 2010 19:52:59 +0000 (21:52 +0200)]
debugutils: Don't consume preroll buffer twice

14 years agopulse: Style fix: use g_strdup() instead of printf()ing a simple string
Benjamin Otte [Tue, 1 Jun 2010 19:32:11 +0000 (21:32 +0200)]
pulse: Style fix: use g_strdup() instead of printf()ing a simple string

14 years agodebugutils: Replace md5 implementation with glib's
Benjamin Otte [Thu, 27 May 2010 14:07:31 +0000 (16:07 +0200)]
debugutils: Replace md5 implementation with glib's

https://bugzilla.gnome.org/show_bug.cgi?id=619824

14 years agoavimux: clean up code for avi header using a bytewriter
Mark Nauwelaerts [Sat, 22 May 2010 09:55:37 +0000 (11:55 +0200)]
avimux: clean up code for avi header using a bytewriter

https://bugzilla.gnome.org/show_bug.cgi?id=619293

14 years agopulsesink: optimize communication with PulseAudio using pa_stream_begin_write
Pierre-Louis Bossart [Tue, 1 Jun 2010 23:54:41 +0000 (18:54 -0500)]
pulsesink: optimize communication with PulseAudio using pa_stream_begin_write

14 years agopulsesink: Post provide-clock message on the bus if the clock appears/disappears
Sebastian Dröge [Wed, 2 Jun 2010 08:52:56 +0000 (10:52 +0200)]
pulsesink: Post provide-clock message on the bus if the clock appears/disappears

Fixes bug #620277.

14 years agoAutomatic update of common submodule
David Schleef [Wed, 2 Jun 2010 06:49:17 +0000 (23:49 -0700)]
Automatic update of common submodule

From 17f89e5 to 47683c1

14 years agoAutomatic update of common submodule
David Schleef [Wed, 2 Jun 2010 05:54:49 +0000 (22:54 -0700)]
Automatic update of common submodule

From cdff0fb to 17f89e5

14 years agovideomixer: filter caps returned from downstream with our pad template.
Edward Hervey [Tue, 1 Jun 2010 18:45:29 +0000 (20:45 +0200)]
videomixer: filter caps returned from downstream with our pad template.

14 years agomatroskamux: Remove more unneeded warnings
Zaheer Abbas Merali [Tue, 1 Jun 2010 15:56:32 +0000 (16:56 +0100)]
matroskamux: Remove more unneeded warnings

14 years agomatroskamux: remove unneeded warning
Zaheer Abbas Merali [Tue, 1 Jun 2010 15:54:03 +0000 (16:54 +0100)]
matroskamux: remove unneeded warning

14 years agomatroskamux: remove unneeded debug statement
Zaheer Abbas Merali [Tue, 1 Jun 2010 15:49:14 +0000 (16:49 +0100)]
matroskamux: remove unneeded debug statement

14 years agomatroskamux: change is-live property to indexed
Zaheer Abbas Merali [Tue, 1 Jun 2010 15:24:53 +0000 (16:24 +0100)]
matroskamux: change is-live property to indexed

14 years agomatroska: use the uint64 scaling functions
Zaheer Abbas Merali [Sun, 23 May 2010 12:56:16 +0000 (13:56 +0100)]
matroska: use the uint64 scaling functions

In demuxer and muxer use the gst_util_uint64 scaling functions rather than
standard integer division. Add warnings (to be changed to debug) for debugging
the timestamp and duration.

14 years agomatroskamux: set delta unit on all buffers except cluster start ones
Zaheer Abbas Merali [Fri, 21 May 2010 13:35:34 +0000 (14:35 +0100)]
matroskamux: set delta unit on all buffers except cluster start ones

14 years agomatroskamux: store caps and set on buffers rather than using pad caps
Zaheer Abbas Merali [Fri, 21 May 2010 12:38:11 +0000 (13:38 +0100)]
matroskamux: store caps and set on buffers rather than using pad caps

14 years agomatroskamux: make sure pads caps are set before any buffers pushed.
Zaheer Abbas Merali [Fri, 21 May 2010 12:25:24 +0000 (13:25 +0100)]
matroskamux: make sure pads caps are set before any buffers pushed.

14 years agomatroskamux: add streamheaders
Zaheer Abbas Merali [Fri, 21 May 2010 12:14:04 +0000 (13:14 +0100)]
matroskamux: add streamheaders

14 years agomatroskamux: no need to set cache twice
Zaheer Abbas Merali [Fri, 21 May 2010 11:23:08 +0000 (12:23 +0100)]
matroskamux: no need to set cache twice

14 years agoDo not create a SeekHeader, Cues, .. when doing live
Xavier Queralt [Thu, 20 May 2010 23:59:53 +0000 (01:59 +0200)]
Do not create a SeekHeader, Cues, .. when doing live

14 years agoAdd is-live property
Xavier Queralt [Thu, 20 May 2010 21:39:59 +0000 (23:39 +0200)]
Add is-live property

14 years agojpegdec: fix variable init
Mark Nauwelaerts [Tue, 1 Jun 2010 11:22:26 +0000 (13:22 +0200)]
jpegdec: fix variable init

14 years agomatroskademux: improve reverse playback
Mark Nauwelaerts [Fri, 28 May 2010 14:37:32 +0000 (16:37 +0200)]
matroskademux: improve reverse playback

Slightly modify approach to also handle cases where cue entries do not reliably
lead to initial keyframes.

Fixes #619817.

14 years agodeinterlace: avoid gtk-doc confusing comments
Mark Nauwelaerts [Mon, 24 May 2010 14:02:58 +0000 (16:02 +0200)]
deinterlace: avoid gtk-doc confusing comments

14 years agomatroskamux: adjust unit test to modified behaviour
Mark Nauwelaerts [Fri, 21 May 2010 09:21:58 +0000 (11:21 +0200)]
matroskamux: adjust unit test to modified behaviour

14 years agomatroskamux: use write caching also when writing buffer data
Mark Nauwelaerts [Thu, 20 May 2010 12:33:41 +0000 (14:33 +0200)]
matroskamux: use write caching also when writing buffer data

Specifically, this reduces pushing several small buffers for each
data buffer and also avoids a seek for each buffer altogether
(though a seek is still needed for each cluster).

Fixes #619273.

14 years agomatroskamux: fix ebml write caching with bytewriter implementation
Mark Nauwelaerts [Thu, 20 May 2010 12:23:07 +0000 (14:23 +0200)]
matroskamux: fix ebml write caching with bytewriter implementation

Also cache a bit more during header writing.

Fixes #619273.

14 years agomatroskamux: use consistent debug category name for ebmlwrite
Mark Nauwelaerts [Thu, 20 May 2010 12:08:42 +0000 (14:08 +0200)]
matroskamux: use consistent debug category name for ebmlwrite

14 years agomatroskademux: use bytereader based GstEbmlRead as a helper
Mark Nauwelaerts [Tue, 18 May 2010 12:44:15 +0000 (14:44 +0200)]
matroskademux: use bytereader based GstEbmlRead as a helper

... rather than basing on it by inheritance.
Also use more common code for push and pull mode.

Fixes #619198.
Fixes #611117.

14 years agomatroskamux: _get_pad_template result needs no unref
Mark Nauwelaerts [Tue, 1 Jun 2010 13:47:32 +0000 (15:47 +0200)]
matroskamux: _get_pad_template result needs no unref

14 years agopngenc: Support 8 bit grayscale
Thiago Santos [Tue, 18 May 2010 22:42:58 +0000 (19:42 -0300)]
pngenc: Support 8 bit grayscale

Adds support to 8 bit grayscale input

14 years agojpegdec: Adds 8bit grayscale support
Thiago Santos [Tue, 18 May 2010 17:46:54 +0000 (14:46 -0300)]
jpegdec: Adds 8bit grayscale support

Adds decoding support for jpeg images in 8 bit grayscale format.

14 years agojpegenc: Accept grayscale as input
Thiago Santos [Tue, 18 May 2010 04:57:14 +0000 (01:57 -0300)]
jpegenc: Accept grayscale as input

Adds video/x-raw-grayscale (8 bit) support to jpegenc

14 years agovideomixer: Implement sinkpad GetCapsFunction.
Edward Hervey [Mon, 31 May 2010 11:30:05 +0000 (13:30 +0200)]
videomixer: Implement sinkpad GetCapsFunction.

This allows returning only the formats, width, height, framerate
and pixel-aspect-ratio that downstream can support.

https://bugzilla.gnome.org/show_bug.cgi?id=620148

14 years agomatroskademux: Don't compare running times with stream times when doing QoS
Sebastian Dröge [Mon, 31 May 2010 05:49:21 +0000 (07:49 +0200)]
matroskademux: Don't compare running times with stream times when doing QoS

14 years agodeinterlace: Don't reconfigure the caps when changing properties
Sebastian Dröge [Thu, 27 May 2010 19:06:43 +0000 (21:06 +0200)]
deinterlace: Don't reconfigure the caps when changing properties

Fixes bug #619848.

14 years agoalpha: Add property to allow passthrough mode
Sebastian Dröge [Wed, 26 May 2010 11:13:44 +0000 (13:13 +0200)]
alpha: Add property to allow passthrough mode

This passthrough mode is used if the alpha method is "set"
and the alpha value is 1.0.

Fixes bug #617512.

14 years agospectrum: support 24-bit width
Alexander Kojevnikov [Tue, 25 May 2010 05:16:06 +0000 (15:16 +1000)]
spectrum: support 24-bit width

Fixes #619045

14 years agospectrum: support arbitrary bit depth
Alexander Kojevnikov [Mon, 24 May 2010 11:50:58 +0000 (21:50 +1000)]
spectrum: support arbitrary bit depth

Partially fixes #619045

14 years agomatroskademux: fix deadlock introduced by video keyframe QoS
Philip Jägenstedt [Tue, 25 May 2010 03:36:46 +0000 (05:36 +0200)]
matroskademux: fix deadlock introduced by video keyframe QoS

14 years agomatroskademux: skip buffers before a late keyframe (QoS)
Philip Jägenstedt [Sun, 23 May 2010 07:32:08 +0000 (09:32 +0200)]
matroskademux: skip buffers before a late keyframe (QoS)

Before, vp8dec had no option but to decode all frames even if some/all
of them would be late. With this change, performance when keyframes are
frequent is helped a great deal. On my Thinkpad X60s, decoding a 20 s
1080p sunflower encode with keyframes every 10 frames went from taking
42 s with 5 frames shown to 21 s with 15 frames shown (still slow
enough to count by hand). When keyframes are more sparse, you will
still be able to catch up eventually, but the results won't be as
noticable.

14 years agovideomixer: Don't mix input with different pixel aspect ratios
Sebastian Dröge [Fri, 14 May 2010 15:57:59 +0000 (17:57 +0200)]
videomixer: Don't mix input with different pixel aspect ratios

Fixes bug #618530.

14 years agodeinterlace: Add MMX/3DNow implementations of greedyh for UYVY
Sebastian Dröge [Mon, 17 May 2010 17:54:22 +0000 (19:54 +0200)]
deinterlace: Add MMX/3DNow implementations of greedyh for UYVY

14 years agodeinterlace: Fix UYVY implementation of greedyh to be actually used
Sebastian Dröge [Mon, 17 May 2010 17:16:43 +0000 (19:16 +0200)]
deinterlace: Fix UYVY implementation of greedyh to be actually used

14 years agogdkpixbuf: Add a gdkpixbuf3 plugin that uses gdkpixbuf3
Sebastian Dröge [Tue, 11 May 2010 09:43:07 +0000 (11:43 +0200)]
gdkpixbuf: Add a gdkpixbuf3 plugin that uses gdkpixbuf3

14 years agowin32: add more generated marshal and enumtype files to win32-update
Tim-Philipp Müller [Tue, 1 Jun 2010 09:06:10 +0000 (10:06 +0100)]
win32: add more generated marshal and enumtype files to win32-update

14 years agoRevert "matroska: add temporary webm typefinder"
Tim-Philipp Müller [Tue, 1 Jun 2010 08:27:00 +0000 (09:27 +0100)]
Revert "matroska: add temporary webm typefinder"

This reverts commit d148ec0ad2053abb0c38fc681a8953292985388f.

We depend on -base git now, which has a webm typefinder in the usual
place.

14 years agoRevert "avimux, flvmux, matroskamux: don't crash if tags arrive on multiple input...
Tim-Philipp Müller [Tue, 1 Jun 2010 08:26:11 +0000 (09:26 +0100)]
Revert "avimux, flvmux, matroskamux: don't crash if tags arrive on multiple input pads at the same time"

This reverts commit 6a9983cd20c48b96396229b3f94d0254a05ddf48.

Rely on locking done in GstTagSetter in core git.

14 years agoconfigure: require core/base git
Tim-Philipp Müller [Tue, 1 Jun 2010 08:23:18 +0000 (09:23 +0100)]
configure: require core/base git

For WebM typefinding and GstTagsetter fixes.

14 years agoBack to development
Tim-Philipp Müller [Tue, 1 Jun 2010 08:17:52 +0000 (09:17 +0100)]
Back to development

14 years agoRelease 0.10.23
Tim-Philipp Müller [Sun, 30 May 2010 13:03:53 +0000 (14:03 +0100)]
Release 0.10.23

14 years agoUpdate .po files
Tim-Philipp Müller [Sun, 30 May 2010 13:02:04 +0000 (14:02 +0100)]
Update .po files

14 years agoflvdemux: Fix position query
Sebastian Dröge [Sat, 29 May 2010 08:23:48 +0000 (10:23 +0200)]
flvdemux: Fix position query

14 years agodocs: remove unnecessary videorate element from webmmux example pipeline
Tim-Philipp Müller [Fri, 28 May 2010 14:14:07 +0000 (15:14 +0100)]
docs: remove unnecessary videorate element from webmmux example pipeline

14 years agojpegenc: Keep variables in sane state after _reset
Thiago Santos [Fri, 28 May 2010 13:43:36 +0000 (10:43 -0300)]
jpegenc: Keep variables in sane state after _reset

When reseting, keep 'row' variables at a sane state after
freeing to avoid it being freed again on _resync realloc
when the element is reused.

Fixes #619943

14 years agovideobox: Fix floating point to integer conversion for the alpha values
Sebastian Dröge [Thu, 27 May 2010 16:08:17 +0000 (18:08 +0200)]
videobox: Fix floating point to integer conversion for the alpha values

Fixes bug #619835.

14 years ago0.10.22.3 pre-release
Tim-Philipp Müller [Wed, 26 May 2010 07:54:33 +0000 (08:54 +0100)]
0.10.22.3 pre-release

14 years agopo: update translations
Tim-Philipp Müller [Tue, 25 May 2010 23:33:59 +0000 (00:33 +0100)]
po: update translations

14 years agowavparse: handle truncated input data at EOS in pull mode
Mark Nauwelaerts [Tue, 25 May 2010 13:34:11 +0000 (15:34 +0200)]
wavparse: handle truncated input data at EOS in pull mode

Fixes #617733.

14 years agoAutomatic update of common submodule
Tim-Philipp Müller [Wed, 26 May 2010 10:55:13 +0000 (11:55 +0100)]
Automatic update of common submodule

From 357b0db to fd7ca04

14 years agoqtdemux: Round timestamp up when scaling to mov format
Robert Swain [Tue, 25 May 2010 19:14:05 +0000 (21:14 +0200)]
qtdemux: Round timestamp up when scaling to mov format

Fix timestamp rounding to allow the correct index to be located.

The issue was that scaling from GStreamer time format to mov time format was
rounding down causing the timestamp of the newsegment event received after a
flushing keyframe seek to find the sample index before the one it should
causing further backward seeking to the keyframe prior until no rounding error
occurred.

Rounding up when scaling to mov format has the desired effect, and it is
not clear whether just the _round () variant would be sufficient.

Fixes bug #619105

14 years agoavimux, flvmux, matroskamux: don't crash if tags arrive on multiple input pads at...
Tim-Philipp Müller [Mon, 24 May 2010 16:26:42 +0000 (17:26 +0100)]
avimux, flvmux, matroskamux: don't crash if tags arrive on multiple input pads at the same time

This is a temporary fix for the release only.

Fixes #619533.

14 years agortptheora: remove delivery-method from caps
Wim Taymans [Tue, 25 May 2010 15:05:12 +0000 (17:05 +0200)]
rtptheora: remove delivery-method from caps

We can accept all delivery methods so don't advertise anything on the caps or
parse anything, we will handle whatever we receive.

Fixes #618940

14 years agomatroska: add temporary webm typefinder
Tim-Philipp Müller [Tue, 25 May 2010 14:40:01 +0000 (15:40 +0100)]
matroska: add temporary webm typefinder

Add webm typefinder just for the release, so webm works for
people whose distros don't patch gst-plugins-base as well.
We'll remove this again after the release.

14 years agodocs: add some pipeline examples to webmmux docs
Tim-Philipp Müller [Sun, 23 May 2010 10:17:27 +0000 (11:17 +0100)]
docs: add some pipeline examples to webmmux docs

14 years agodocs: add webmmux to docs
Tim-Philipp Müller [Fri, 21 May 2010 11:27:07 +0000 (12:27 +0100)]
docs: add webmmux to docs

14 years agomatroska: fix up plugin and element descriptions a bit
Tim-Philipp Müller [Fri, 21 May 2010 12:01:30 +0000 (13:01 +0100)]
matroska: fix up plugin and element descriptions a bit

14 years agomatroska: move webmmux into own source files
Tim-Philipp Müller [Fri, 21 May 2010 11:47:03 +0000 (12:47 +0100)]
matroska: move webmmux into own source files

Makes things easier for gtk-doc.

14 years agoUpdate spec file with latest changes
Christian Schaller [Fri, 21 May 2010 07:26:05 +0000 (12:26 +0500)]
Update spec file with latest changes

14 years agomatroska: Remove the doctype enum, it's not needed anymore
Sebastian Dröge [Thu, 20 May 2010 18:01:58 +0000 (20:01 +0200)]
matroska: Remove the doctype enum, it's not needed anymore

14 years agowebmmux: Add new webmmux element that only supports muxing of WebM
Sebastian Dröge [Thu, 20 May 2010 17:57:14 +0000 (19:57 +0200)]
webmmux: Add new webmmux element that only supports muxing of WebM

...and remove the doctype property from matroskamux again.

14 years agomatroskamux: unit test checks version 1 files
Mark Nauwelaerts [Thu, 20 May 2010 15:31:59 +0000 (17:31 +0200)]
matroskamux: unit test checks version 1 files

14 years agospeex: fix latency query
Tristan Matthews [Tue, 18 May 2010 19:27:06 +0000 (15:27 -0400)]
speex: fix latency query

  Speex should report 30 ms latency for narrowband mode, 34 otherwise.
  Fixes #619018

14 years agoebmlread: rm floatcast.h include (not used)
Philip [Tue, 18 May 2010 13:04:32 +0000 (21:04 +0800)]
ebmlread: rm floatcast.h include (not used)

14 years agomatroskamux: bump default doctype version to 2
Philip Jägenstedt [Mon, 17 May 2010 03:36:00 +0000 (05:36 +0200)]
matroskamux: bump default doctype version to 2

In this day and age this should be safe. There's otherwise a risk people
will be creating unneccessarily big WebM files as they can't use
SimpleBlock in v1.

14 years agomatroska: handle matroska and webm doctype versions equally
Philip Jägenstedt [Mon, 17 May 2010 03:27:44 +0000 (05:27 +0200)]
matroska: handle matroska and webm doctype versions equally

The original plan was to let WebM v1 be the same as Matroska v2 (with
extra constraints), but for simplicity it was decided to handle the
versions equally, such that e.g. SimpleBlock is only allowed in WebM v2.

14 years agomatroskademux: Verify lace size in _parse_blockgroup_or_simpleblock
Philip Jägenstedt [Thu, 13 May 2010 10:10:54 +0000 (12:10 +0200)]
matroskademux: Verify lace size in _parse_blockgroup_or_simpleblock

Failure to do this for corrupt input can cause a subbuffer bigger
than the actual buffer to be created, quickly leading to segfault.
Test case:
bug_s222005751_r0.001____memcpy.webm

14 years agoebml: crude hack to avoid crashing on unexpected metadata
Philip Jägenstedt [Thu, 13 May 2010 08:23:10 +0000 (10:23 +0200)]
ebml: crude hack to avoid crashing on unexpected metadata

The comment says this cannot happen, but it did and I don't know
why. This is not the correct fix, needs investigation. Test case:
bug_s555010094_r0.0005:0.008____IA__g_assertion_message_expr.webm

14 years agoebml: don't modify out str if returning an error in _read_ascii
Philip Jägenstedt [Thu, 13 May 2010 07:18:56 +0000 (09:18 +0200)]
ebml: don't modify out str if returning an error in _read_ascii

This is a regression from ASCII validation changes. Test case:
bug_s66876390_r0.001____malloc_printerr.webm

14 years agoebml: Validate 7-bit ASCII in gst_ebml_read_ascii
Philip Jägenstedt [Wed, 12 May 2010 11:16:28 +0000 (13:16 +0200)]
ebml: Validate 7-bit ASCII in gst_ebml_read_ascii

This was triggering an UTF-8 assertion in gst_caps_set_simple for
corrupt files with garbage as codec id. Test case:
gstreamer_error_trying_to_set_invalid_utf8_as_codec_id.webm

Old gst_ebml_read_ascii renamed to gst_ebml_read_string, also used by
gst_ebml_read_utf8. Unlike for UTF-8, failure to validate is an error,
as gst_ebml_read_ascii is used for reading doctype and codec id and we
might just as well give up early in those cases.

14 years agomatroskademux: Ignore unexpected CodecState
Philip Jägenstedt [Wed, 12 May 2010 12:30:18 +0000 (14:30 +0200)]
matroskademux: Ignore unexpected CodecState

Because GstMatroskaTrackContext *stream is set up in the first
SimpleBlock or Block, a rogue CodecState otherwise causes a segfault on
derefencing the NULL pointer. Test case:
bug_s5506167_r0.001____gst_matroska_demux_parse_blockgroup_or_simpleblock.webm