platform/upstream/libav.git
12 years agolavfi: reorder AVFilterBuffer fields.
Anton Khirnov [Thu, 21 Jun 2012 09:48:58 +0000 (11:48 +0200)]
lavfi: reorder AVFilterBuffer fields.

Place related fields together, remove holes.

12 years agolavfi: remove disabled FF_API_FILTERS_PUBLIC cruft
Anton Khirnov [Sun, 17 Jun 2012 10:36:52 +0000 (12:36 +0200)]
lavfi: remove disabled FF_API_FILTERS_PUBLIC cruft

12 years agolavfi: remove disabled FF_API_DEFAULT_CONFIG_OUTPUT_LINK cruft
Anton Khirnov [Sun, 17 Jun 2012 10:36:52 +0000 (12:36 +0200)]
lavfi: remove disabled FF_API_DEFAULT_CONFIG_OUTPUT_LINK cruft

12 years agolavfi: use proper FF_API guards for different deprecated functions
Anton Khirnov [Tue, 26 Jun 2012 11:17:31 +0000 (13:17 +0200)]
lavfi: use proper FF_API guards for different deprecated functions

12 years agolavfi: remove disabled FF_API_VSRC_BUFFER_ADD_FRAME cruft
Anton Khirnov [Sun, 17 Jun 2012 10:36:52 +0000 (12:36 +0200)]
lavfi: remove disabled FF_API_VSRC_BUFFER_ADD_FRAME cruft

12 years agolavfi: remove disabled FF_API_SAMPLERATE64 cruft
Anton Khirnov [Sun, 17 Jun 2012 10:36:52 +0000 (12:36 +0200)]
lavfi: remove disabled FF_API_SAMPLERATE64 cruft

12 years agolavfi: remove disabled FF_API_GRAPH_AVCLASS cruft
Anton Khirnov [Sun, 17 Jun 2012 10:36:52 +0000 (12:36 +0200)]
lavfi: remove disabled FF_API_GRAPH_AVCLASS cruft

12 years agolavfi: remove 'opaque' parameter from AVFilter.init()
Anton Khirnov [Thu, 21 Jun 2012 05:55:56 +0000 (07:55 +0200)]
lavfi: remove 'opaque' parameter from AVFilter.init()

It is not used in any filters currently and is inherently evil. If
passing binary data to filters is required in the future, it should be
done with some AVOptions-based system.

12 years agomov: do not try to read total disc/track number if data atom is too short.
Carl Eugen Hoyos [Thu, 17 Nov 2011 13:12:34 +0000 (14:12 +0100)]
mov: do not try to read total disc/track number if data atom is too short.

Fixes bug 308.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agoavconv: fix -force_key_frames
Anton Khirnov [Fri, 22 Jun 2012 12:36:27 +0000 (14:36 +0200)]
avconv: fix -force_key_frames

parse_forced_keyframes() relies in encoder timebase being set, so call
it from transcode_init() after it is known.

12 years agodxva2_h264: fix signaling of mbaff frames
Hendrik Leppkes [Mon, 27 Feb 2012 16:21:16 +0000 (17:21 +0100)]
dxva2_h264: fix signaling of mbaff frames

The MBAFF flag may only be signaled if we're actually dealing with
a full frame, and not singular fields, as it can happen in mixed content.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agox86: fft: elf64: fix PIC build
Mans Rullgard [Mon, 25 Jun 2012 14:25:55 +0000 (15:25 +0100)]
x86: fft: elf64: fix PIC build

In a 64-bit PIC build, external functions must be called
through the PLT.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agolog: Include io.h on windows
Ronald S. Bultje [Sun, 24 Jun 2012 17:57:35 +0000 (20:57 +0300)]
log: Include io.h on windows

This is required for isatty, which exists on MSVC and is found by
configure, but is provided by io.h instead of unistd.h.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agolavr: x86: merge some branches
Justin Ruggles [Mon, 25 Jun 2012 00:37:25 +0000 (20:37 -0400)]
lavr: x86: merge some branches

12 years agox86: cpu: whitespace (mostly) cosmetics
Mans Rullgard [Mon, 25 Jun 2012 12:43:32 +0000 (13:43 +0100)]
x86: cpu: whitespace (mostly) cosmetics

This adds whitespace around operators, aligns line continuation
backslashes, and breaks long lines.  Also fixes an ifdef halfway
through a statement.  The one line of duplication this saved is
not worth the ugliness.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agox86: fft: win64: fix stack alignment for memcpy() call
Mans Rullgard [Mon, 25 Jun 2012 13:33:24 +0000 (14:33 +0100)]
x86: fft: win64: fix stack alignment for memcpy() call

12 years agowtv: Check the return value from gmtime
Martin Storsjö [Sat, 23 Jun 2012 18:46:26 +0000 (21:46 +0300)]
wtv: Check the return value from gmtime

On MSVC, gmtime returns NULL for values outside of their supported
range (and these show up in our fate test). This doesn't seem
to affect the actual fate test result.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agox86: fft: convert sse inline asm to yasm
Mans Rullgard [Sat, 23 Jun 2012 18:08:11 +0000 (19:08 +0100)]
x86: fft: convert sse inline asm to yasm

12 years agox86: place some inline asm under #if HAVE_INLINE_ASM
Ronald S. Bultje [Fri, 22 Jun 2012 17:46:52 +0000 (18:46 +0100)]
x86: place some inline asm under #if HAVE_INLINE_ASM

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agolog: Only include unistd.h if configure found it
Martin Storsjö [Sat, 23 Jun 2012 16:21:35 +0000 (19:21 +0300)]
log: Only include unistd.h if configure found it

MSVC has isatty (in io.h), but not unistd.h. (isatty isn't called
at all for windows, since there's a special case block for that.)

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agoape: create audio stream before reading tags.
Anton Khirnov [Fri, 22 Jun 2012 11:08:08 +0000 (13:08 +0200)]
ape: create audio stream before reading tags.

Tags may contain attached picture, which will be exported as video
streams.

This ensures that the audio stream is always the first.

12 years agomov: make a length variable larger.
Anton Khirnov [Thu, 21 Jun 2012 16:13:34 +0000 (18:13 +0200)]
mov: make a length variable larger.

Right now, it's uint16_t, but for itunes metadata a 32bit number is
stored in it.

12 years agoimage2: Add "start_number" private option to the demuxer
Mashiat Sarker Shakkhar [Sat, 23 Jun 2012 03:57:04 +0000 (09:57 +0600)]
image2: Add "start_number" private option to the demuxer

Currently if a pattern is given we look for up to the fifth file name in
the sequence. This option sets that limit to an arbitrary number.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agoimage2: Add "start_number" private option to the muxer
Mashiat Sarker Shakkhar [Sat, 23 Jun 2012 03:53:43 +0000 (09:53 +0600)]
image2: Add "start_number" private option to the muxer

This adds the capability to start counting file number from an arbitrary
integer.

This includes a few lines of trivial code from FFmpeg codebase.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
12 years agoavconv: remove a forgotten debugging printf.
Anton Khirnov [Sat, 23 Jun 2012 14:10:08 +0000 (16:10 +0200)]
avconv: remove a forgotten debugging printf.

12 years agoavconv: use more descriptive names for hardcoded filters.
Anton Khirnov [Sat, 23 Jun 2012 14:08:24 +0000 (16:08 +0200)]
avconv: use more descriptive names for hardcoded filters.

Also ensure that all such filters get a non-NULL name. Should fix FATE
failures on some architectures after
58b049f2fa4f192b00baadb5f1f32ca366f936ea.

12 years agoavconv: remove redundant handling of async.
Anton Khirnov [Sat, 23 Jun 2012 14:02:45 +0000 (16:02 +0200)]
avconv: remove redundant handling of async.

Because of a mistake during merging the code for simple and complex
filtergraphs, -async inserts an asyncts filter both on input and output.
Remove the output hunk.

12 years agodoc/filters: fix typo.
Anton Khirnov [Fri, 22 Jun 2012 21:01:03 +0000 (23:01 +0200)]
doc/filters: fix typo.

12 years agoh264: use asm cabac reader under a generic condition
Mans Rullgard [Fri, 22 Jun 2012 17:52:27 +0000 (18:52 +0100)]
h264: use asm cabac reader under a generic condition

This removes a dependency on implementation details from generic
code and allows easy addition of the equivalent optimisation for
other architectures than x86.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agox86: Only use optimizations with cmov if the CPU supports the instruction
Diego Biurrun [Tue, 19 Jun 2012 10:55:10 +0000 (12:55 +0200)]
x86: Only use optimizations with cmov if the CPU supports the instruction

12 years agox86: Add CPU flag for the i686 cmov instruction
Diego Biurrun [Tue, 19 Jun 2012 20:55:26 +0000 (22:55 +0200)]
x86: Add CPU flag for the i686 cmov instruction

12 years agox86: remove unused inline asm macros from dsputil_mmx.h
Mans Rullgard [Fri, 22 Jun 2012 21:11:02 +0000 (22:11 +0100)]
x86: remove unused inline asm macros from dsputil_mmx.h

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agox86: move some inline asm macros to the only places they are used
Mans Rullgard [Fri, 22 Jun 2012 21:10:31 +0000 (22:10 +0100)]
x86: move some inline asm macros to the only places they are used

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agolavfi: Add the af_channelmap audio channel mapping filter.
Alex Converse [Sat, 9 Jun 2012 01:42:53 +0000 (18:42 -0700)]
lavfi: Add the af_channelmap audio channel mapping filter.

Inspired by MPlayer's af_channels filter and SoX's remix effect.

12 years agolavfi: add join audio filter.
Anton Khirnov [Mon, 28 May 2012 05:28:58 +0000 (07:28 +0200)]
lavfi: add join audio filter.

It joins multiple input streams into one multi-channel output.

12 years agolavfi: allow audio filters to request a given number of samples.
Anton Khirnov [Sun, 27 May 2012 12:18:49 +0000 (14:18 +0200)]
lavfi: allow audio filters to request a given number of samples.

This makes synchronization simpler for filters with multiple inputs.

12 years agolavfi: support automatically inserting the fifo filter when needed.
Anton Khirnov [Wed, 16 May 2012 07:19:46 +0000 (09:19 +0200)]
lavfi: support automatically inserting the fifo filter when needed.

This breaks libavfilter ABI.

12 years agolavfi/audio: eliminate ff_default_filter_samples().
Anton Khirnov [Sun, 17 Jun 2012 08:09:54 +0000 (10:09 +0200)]
lavfi/audio: eliminate ff_default_filter_samples().

It currently does the following:
1) get a zeroed audio buffer
2) copy some properties (but not the data) of the input buffer to it
3) pass this buffer to the output filter
This looks useless and is indeed not used by any filters, therefore
delete it.

Make ff_null_filter_samples() (just pass the buffer to the next filter)
the new default.

12 years agolibspeexenc: add supported sample rates and channel layouts.
Anton Khirnov [Thu, 21 Jun 2012 07:19:23 +0000 (09:19 +0200)]
libspeexenc: add supported sample rates and channel layouts.

12 years agoReplace usleep() calls with av_usleep()
Mans Rullgard [Thu, 21 Jun 2012 19:31:44 +0000 (20:31 +0100)]
Replace usleep() calls with av_usleep()

This reduces the dependency on unistd.h which is not available
on all systems.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agolavu: add av_usleep() function
Mans Rullgard [Thu, 21 Jun 2012 16:00:25 +0000 (17:00 +0100)]
lavu: add av_usleep() function

This function implements a delay using the first available
of the following functions:

- nanosleep()
- usleep()
- Sleep() (Windows)

The conditional #includes in time.c are simplified by including
unistd.h and windows.h whenever they are available rather than
having these lines triggered by specific functions.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agoutvideo: mark interlaced frames as such
Hendrik Leppkes [Thu, 21 Jun 2012 21:11:24 +0000 (21:11 +0000)]
utvideo: mark interlaced frames as such

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
12 years agoutvideo: Fix interlaced prediction for RGB utvideo.
Carl Eugen Hoyos [Thu, 21 Jun 2012 21:11:23 +0000 (21:11 +0000)]
utvideo: Fix interlaced prediction for RGB utvideo.

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
12 years agocosmetics: do not use full path for local headers
Diego Biurrun [Thu, 21 Jun 2012 09:44:53 +0000 (11:44 +0200)]
cosmetics: do not use full path for local headers

12 years agolavu/file: include unistd.h only when available
Mans Rullgard [Thu, 21 Jun 2012 15:54:02 +0000 (16:54 +0100)]
lavu/file: include unistd.h only when available

The unistd.h header is only needed for the close() declaration.
If this header is not available, the close() declaration may be
provided by another header, e.g. io.h.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agoconfigure: check for unistd.h
Mans Rullgard [Thu, 21 Jun 2012 15:34:10 +0000 (16:34 +0100)]
configure: check for unistd.h

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agolog: include unistd.h only when needed
Mans Rullgard [Thu, 21 Jun 2012 15:33:36 +0000 (16:33 +0100)]
log: include unistd.h only when needed

The only symbol this file uses from unistd.h is isatty().  By
including the header only when this function is used, the file
can be built on systems without unistd.h (which presumably also
lack isatty).

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agolavf: include libavutil/time.h instead of redeclaring av_gettime()
Mans Rullgard [Thu, 21 Jun 2012 15:36:15 +0000 (16:36 +0100)]
lavf: include libavutil/time.h instead of redeclaring av_gettime()

This avoids some warnings about redundant declarations.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agofix hardcoded tables compililation caused by missing math constants
Janne Grunau [Thu, 21 Jun 2012 15:59:56 +0000 (17:59 +0200)]
fix hardcoded tables compililation caused by missing math constants

Add -D_XOPEN_SOURCE=600 to host cflags to make the constants in math.h
available. Include math.h where necessary and remove redundant M_PI
defines.

12 years agolavf: Make codec_tag arrays constant
Ronald S. Bultje [Wed, 20 Jun 2012 17:25:00 +0000 (10:25 -0700)]
lavf: Make codec_tag arrays constant

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agotwinvq: give massive struct a name.
Ronald S. Bultje [Wed, 20 Jun 2012 17:11:43 +0000 (10:11 -0700)]
twinvq: give massive struct a name.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agolavf, lavu: version bumps and APIchanges for av_gettime() move
Mans Rullgard [Wed, 20 Jun 2012 16:28:34 +0000 (17:28 +0100)]
lavf, lavu: version bumps and APIchanges for av_gettime() move

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agolavfi/audio: don't set cur_buf in ff_filter_samples().
Anton Khirnov [Sun, 17 Jun 2012 08:21:49 +0000 (10:21 +0200)]
lavfi/audio: don't set cur_buf in ff_filter_samples().

It's redundant, since the input buffer is passed as a parameter to the
filter_samples() callback, and can lead to stale pointers remaining on
the link.

12 years agolavfi/fifo: add audio version of the fifo filter.
Anton Khirnov [Wed, 16 May 2012 06:21:47 +0000 (08:21 +0200)]
lavfi/fifo: add audio version of the fifo filter.

12 years agofifo: fix parenthesis placement.
Anton Khirnov [Sun, 27 May 2012 20:45:52 +0000 (22:45 +0200)]
fifo: fix parenthesis placement.

12 years agolavfi: rename vf_fifo.c -> fifo.c
Anton Khirnov [Wed, 16 May 2012 06:15:54 +0000 (08:15 +0200)]
lavfi: rename vf_fifo.c -> fifo.c

It will be used for audio too.

12 years agolavc: remove stats_in from AVCodecContext options table.
Anton Khirnov [Wed, 20 Jun 2012 16:36:14 +0000 (18:36 +0200)]
lavc: remove stats_in from AVCodecContext options table.

Its documentation states that it is allocated/freed by the caller, but
it is declared as an AV_OPT_TYPE_STRING AVOption. Since
367732832faaf1bac4ece37cf7fef8c911e16312 the AVOptions system frees
strings automatically. This can be considered an API break, since it
won't work when the caller doesn't use av_malloc() to allocate the
memory or wants to use the string after closing the codec.

Since there is not much value in this field being an AVOption, the best
solution is to remove it from the options table.

12 years agoMS Screen 1 decoder
Kostya Shishkov [Fri, 15 Jun 2012 18:37:37 +0000 (20:37 +0200)]
MS Screen 1 decoder

12 years agoaacdec: Fix popping channel layouts.
Alex Converse [Wed, 20 Jun 2012 02:06:34 +0000 (19:06 -0700)]
aacdec: Fix popping channel layouts.

'channel_layout' not 'channels' from the stored configuration should go
to AVCodecContext's 'channel_layout'.

12 years agoav_gettime: support Win32 without gettimeofday()
Mans Rullgard [Wed, 20 Jun 2012 12:46:52 +0000 (13:46 +0100)]
av_gettime: support Win32 without gettimeofday()

Based on patch by Ronald S. Bultje.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agoUse av_gettime() in various places
Mans Rullgard [Wed, 20 Jun 2012 00:34:38 +0000 (01:34 +0100)]
Use av_gettime() in various places

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agoMove av_gettime() to libavutil
Mans Rullgard [Wed, 20 Jun 2012 00:24:39 +0000 (01:24 +0100)]
Move av_gettime() to libavutil

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agodct-test: use emms_c() from libavutil instead of duplicating it
Mans Rullgard [Wed, 20 Jun 2012 12:50:45 +0000 (13:50 +0100)]
dct-test: use emms_c() from libavutil instead of duplicating it

12 years agomov: fix operator precedence bug
Ronald S. Bultje [Mon, 18 Jun 2012 19:06:35 +0000 (20:06 +0100)]
mov: fix operator precedence bug

12 years agomathematics.h: remove a couple of math defines
Janne Grunau [Thu, 14 Jun 2012 18:13:00 +0000 (20:13 +0200)]
mathematics.h: remove a couple of math defines

While these defines are not defined by the C standard they are
standardized as X/Open System Interfaces Extension. We use the
appropiate _XOPEN_SOURCE define to make them available. They
seem to be available on all FATE configs since the constants
are used in files where mathematics.h is not included.

12 years agoRemove unnecessary inclusions of [sys/]time.h
Mans Rullgard [Tue, 19 Jun 2012 21:08:35 +0000 (22:08 +0100)]
Remove unnecessary inclusions of [sys/]time.h

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agolavf: remove unnecessary inclusions of unistd.h
Mans Rullgard [Tue, 19 Jun 2012 14:11:34 +0000 (15:11 +0100)]
lavf: remove unnecessary inclusions of unistd.h

These files do not use anything provided by unistd.h.

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agobfin: libswscale: add const where appropriate to fix warnings
Mans Rullgard [Tue, 19 Jun 2012 14:14:33 +0000 (15:14 +0100)]
bfin: libswscale: add const where appropriate to fix warnings

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agobfin: libswscale: remove unnecessary #includes
Mans Rullgard [Tue, 19 Jun 2012 14:12:57 +0000 (15:12 +0100)]
bfin: libswscale: remove unnecessary #includes

Signed-off-by: Mans Rullgard <mans@mansr.com>
12 years agoudp: Properly check for invalid sockets
Martin Storsjö [Tue, 19 Jun 2012 14:59:57 +0000 (17:59 +0300)]
udp: Properly check for invalid sockets

If stdin has been closed, 0 is a valid socket descriptor.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agotcp: Check the return value from getsockopt
Martin Storsjö [Tue, 19 Jun 2012 14:50:38 +0000 (17:50 +0300)]
tcp: Check the return value from getsockopt

Make sure we actually have an error code in ret, in case
getsockopt failed.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agonetwork: Use av_strerror for getting error messages
Martin Storsjö [Tue, 19 Jun 2012 12:37:06 +0000 (15:37 +0300)]
network: Use av_strerror for getting error messages

Also use ff_neterrno() instead of errno directly (which doesn't work
on windows), for getting the error code.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agoudp: Properly print error from getnameinfo
Martin Storsjö [Tue, 19 Jun 2012 12:28:48 +0000 (15:28 +0300)]
udp: Properly print error from getnameinfo

getnameinfo doesn't set errno on failure, it returns an error code,
which should be handled by gai_strerror instead of the normal
strerror.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agommst: Use AVUNERROR() to convert error codes to the right range for strerror
Martin Storsjö [Tue, 19 Jun 2012 12:21:43 +0000 (15:21 +0300)]
mmst: Use AVUNERROR() to convert error codes to the right range for strerror

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agonetwork: Pass pointers of the right type to get/setsockopt/ioctlsocket on windows
Martin Storsjö [Mon, 18 Jun 2012 20:39:30 +0000 (23:39 +0300)]
network: Pass pointers of the right type to get/setsockopt/ioctlsocket on windows

This avoids warnings.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agortmp: Reduce the number of idle posts sent by sleeping 50ms
Samuel Pitoiset [Tue, 19 Jun 2012 11:21:09 +0000 (13:21 +0200)]
rtmp: Reduce the number of idle posts sent by sleeping 50ms

Rtmpt is effectively half duplex - the server can't return any
data unless we send a request (to which the server responds). If
we don't have any data to send currently, and the server didn't
return any data either, wait a little before doing the next request.

This avoids busy looping with idle posts with empty replies, while
waiting for more data from the server.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agoflvdec: remove incomplete, disabled seeking code
Diego Biurrun [Fri, 29 Jul 2011 10:02:54 +0000 (12:02 +0200)]
flvdec: remove incomplete, disabled seeking code

12 years agomem: add support for _aligned_malloc() as found on Windows
Ronald S. Bultje [Mon, 18 Jun 2012 13:37:02 +0000 (14:37 +0100)]
mem: add support for _aligned_malloc() as found on Windows

The check uses check_func_header, since this function is
conditionally available depending on the targeted MSVCRT
version.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agolavc: Extend the documentation for avcodec_init_packet
Simon A. Eugster [Sat, 9 Jun 2012 09:16:12 +0000 (11:16 +0200)]
lavc: Extend the documentation for avcodec_init_packet

Add a note that pkt->data and pkt->size must be initialized.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agoflvdec: remove incomplete, disabled seeking code
Ronald S. Bultje [Mon, 18 Jun 2012 14:22:15 +0000 (15:22 +0100)]
flvdec: remove incomplete, disabled seeking code

12 years agohttp: replace atoll() with strtoll()
Ronald S. Bultje [Mon, 18 Jun 2012 13:02:43 +0000 (14:02 +0100)]
http: replace atoll() with strtoll()

12 years agompegts: remove unused/incomplete/broken seeking code
Ronald S. Bultje [Mon, 18 Jun 2012 12:57:59 +0000 (13:57 +0100)]
mpegts: remove unused/incomplete/broken seeking code

12 years agoaf_amix: allow float planar sample format as input
Justin Ruggles [Sat, 9 Jun 2012 03:59:04 +0000 (23:59 -0400)]
af_amix: allow float planar sample format as input

12 years agoaf_amix: use AVFloatDSPContext.vector_fmac_scalar()
Justin Ruggles [Sat, 9 Jun 2012 02:34:30 +0000 (22:34 -0400)]
af_amix: use AVFloatDSPContext.vector_fmac_scalar()

12 years agofloat_dsp: add x86-optimized functions for vector_fmac_scalar()
Justin Ruggles [Sat, 9 Jun 2012 03:20:59 +0000 (23:20 -0400)]
float_dsp: add x86-optimized functions for vector_fmac_scalar()

12 years agofloat_dsp: Move vector_fmac_scalar() from libavcodec to libavutil
Justin Ruggles [Fri, 8 Jun 2012 17:49:56 +0000 (13:49 -0400)]
float_dsp: Move vector_fmac_scalar() from libavcodec to libavutil

12 years agolavr: Add x86-optimized function for flt to s32 conversion
Justin Ruggles [Fri, 20 Apr 2012 20:09:15 +0000 (16:09 -0400)]
lavr: Add x86-optimized function for flt to s32 conversion

12 years agolavr: Add x86-optimized function for flt to s16 conversion
Justin Ruggles [Tue, 24 Apr 2012 02:22:28 +0000 (22:22 -0400)]
lavr: Add x86-optimized function for flt to s16 conversion

12 years agolavr: Add x86-optimized functions for s32 to flt conversion
Justin Ruggles [Mon, 23 Apr 2012 22:29:58 +0000 (18:29 -0400)]
lavr: Add x86-optimized functions for s32 to flt conversion

12 years agolavr: Add x86-optimized functions for s32 to s16 conversion
Justin Ruggles [Fri, 20 Apr 2012 17:49:53 +0000 (13:49 -0400)]
lavr: Add x86-optimized functions for s32 to s16 conversion

12 years agolavr: Add x86-optimized functions for s16 to flt conversion
Justin Ruggles [Tue, 24 Apr 2012 02:10:26 +0000 (22:10 -0400)]
lavr: Add x86-optimized functions for s16 to flt conversion

12 years agolavr: Add x86-optimized function for s16 to s32 conversion
Justin Ruggles [Fri, 20 Apr 2012 19:48:08 +0000 (15:48 -0400)]
lavr: Add x86-optimized function for s16 to s32 conversion

12 years agortpenc: Support packetizing iLBC
Martin Storsjö [Sun, 17 Jun 2012 14:25:46 +0000 (17:25 +0300)]
rtpenc: Support packetizing iLBC

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agortpdec: Add a depacketizer for iLBC
Martin Storsjö [Sun, 17 Jun 2012 13:12:53 +0000 (16:12 +0300)]
rtpdec: Add a depacketizer for iLBC

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agoImplement the iLBC storage file format
Martin Storsjö [Sun, 17 Jun 2012 12:54:31 +0000 (15:54 +0300)]
Implement the iLBC storage file format

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agomov: Support muxing/demuxing iLBC
Martin Storsjö [Sat, 16 Jun 2012 21:29:26 +0000 (00:29 +0300)]
mov: Support muxing/demuxing iLBC

The packet size, signalled via block_align, has to be passed via
the container.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agoAdd support for iLBC decoding/encoding via the external library libilbc
Martin Storsjö [Fri, 15 Jun 2012 21:42:13 +0000 (00:42 +0300)]
Add support for iLBC decoding/encoding via the external library libilbc

The library is 3-clause BSD licensed.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agortmp: Don't send every flv packet in a separate HTTP request in RTMPT
Samuel Pitoiset [Mon, 18 Jun 2012 12:55:55 +0000 (14:55 +0200)]
rtmp: Don't send every flv packet in a separate HTTP request in RTMPT

Add a new option 'rtmp_flush_interval' that allows specifying the
number of packets to write before sending it off as a HTTP request.

This is mostly relevant for RTMPT - for plain RTMP, it only controls
how often we check the socket for incoming packets, which shouldn't
affect the performance in any noticeable way.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agortsp: Send mode=record instead of mode=receive
Martin Storsjö [Mon, 18 Jun 2012 13:19:33 +0000 (16:19 +0300)]
rtsp: Send mode=record instead of mode=receive

This seems to be the correct mode to send, according to the
original RTSP RFC, and matches the method RECORD which is
sent later when starting to send data.

Darwin Streaming Server works fine with either of them.

Signed-off-by: Martin Storsjö <martin@martin.st>
12 years agodirac: replace compound literal with normal initialiser
Ronald S. Bultje [Mon, 18 Jun 2012 11:57:25 +0000 (12:57 +0100)]
dirac: replace compound literal with normal initialiser

Signed-off-by: Mans Rullgard <mans@mansr.com>