13 years agoflacdec: cosmetics: Add a comment with the data of the smallest FLAC
Justin Ruggles [Thu, 5 Mar 2009 04:40:42 +0000 (04:40 +0000)]
flacdec: cosmetics: Add a comment with the data of the smallest FLAC
frame for future reference.

Originally committed as revision 17822 to svn://

13 years agoflacdec: There is an even smaller FLAC frame size possibility.
Justin Ruggles [Thu, 5 Mar 2009 02:36:17 +0000 (02:36 +0000)]
flacdec: There is an even smaller FLAC frame size possibility.

Originally committed as revision 17821 to svn://

13 years agoflacdec: 10l to me. I miscalculated the smallest FLAC frame. It is 16
Justin Ruggles [Thu, 5 Mar 2009 02:21:35 +0000 (02:21 +0000)]
flacdec: 10l to me. I miscalculated the smallest FLAC frame. It is 16
bytes, not 24.

Originally committed as revision 17820 to svn://

13 years agoflacdec: cosmetics: move 2 statements to immediately after frame
Justin Ruggles [Thu, 5 Mar 2009 01:23:53 +0000 (01:23 +0000)]
flacdec: cosmetics: move 2 statements to immediately after frame

Originally committed as revision 17819 to svn://

13 years agoflacdec: Simplify frame sync by not using the bitstream reader.
Justin Ruggles [Thu, 5 Mar 2009 01:15:38 +0000 (01:15 +0000)]
flacdec: Simplify frame sync by not using the bitstream reader.

Originally committed as revision 17818 to svn://

13 years agoflacdec: Split the metadata_parse() function into 2 separate functions,
Justin Ruggles [Thu, 5 Mar 2009 00:56:09 +0000 (00:56 +0000)]
flacdec: Split the metadata_parse() function into 2 separate functions,
parse_streaminfo() and get_metadata_size().

Originally committed as revision 17817 to svn://

13 years agoflacdec: Add a check for small buffer size. This ensures reading as
Justin Ruggles [Wed, 4 Mar 2009 23:55:10 +0000 (23:55 +0000)]
flacdec: Add a check for small buffer size.  This ensures reading as
much of the frame header as possible without excluding the smallest
possible FLAC frame.  It also fixes a false positive warning message
that was being emitted at the end of decoding.

Originally committed as revision 17816 to svn://

13 years agoflacdec: cosmetics: Use a more descriptive variable name for the number
Justin Ruggles [Wed, 4 Mar 2009 23:24:44 +0000 (23:24 +0000)]
flacdec: cosmetics: Use a more descriptive variable name for the number
of bytes read, instead of reusing 'i'.

Originally committed as revision 17815 to svn://

13 years agoflacdec: Handle 32 bps data when checking allocated size.
Justin Ruggles [Wed, 4 Mar 2009 23:16:30 +0000 (23:16 +0000)]
flacdec: Handle 32 bps data when checking allocated size.

Originally committed as revision 17814 to svn://

13 years agoflacdec: Check curr_bps after wasted bits are subtracted.
Justin Ruggles [Wed, 4 Mar 2009 23:10:40 +0000 (23:10 +0000)]
flacdec: Check curr_bps after wasted bits are subtracted.

Originally committed as revision 17813 to svn://

13 years agoFix typo: "i" -> "in".
Stefano Sabatini [Wed, 4 Mar 2009 21:50:10 +0000 (21:50 +0000)]
Fix typo: "i" -> "in".

Originally committed as revision 17812 to svn://

13 years agoAdd support for ct_type to correctly detect interlaced flag
Ivan Schreter [Wed, 4 Mar 2009 19:37:19 +0000 (19:37 +0000)]
Add support for ct_type to correctly detect interlaced flag

Originally committed as revision 17811 to svn://

13 years agoRewrite of rgb15to32 and rgb16to32 using fewer asm instructions and setting alpha...
Cédric Schieli [Wed, 4 Mar 2009 19:12:49 +0000 (19:12 +0000)]
Rewrite of rgb15to32 and rgb16to32 using fewer asm instructions and setting alpha channel to 0xFF

Originally committed as revision 28808 to svn://

13 years agoAdd "deprecated old scaler removed"
Michael Niedermayer [Wed, 4 Mar 2009 18:43:20 +0000 (18:43 +0000)]
Add "deprecated old scaler removed"

Originally committed as revision 17810 to svn://

13 years agoRemove the swscale dummy option, most (all?) FATE configs should be updated
Reimar Döffinger [Wed, 4 Mar 2009 18:41:47 +0000 (18:41 +0000)]
Remove the swscale dummy option, most (all?) FATE configs should be updated
to work without it.

Originally committed as revision 17809 to svn://

13 years ago100l, interplayvideo uses 8x8 blocks, so call the 8x8 put_pixels_tab function
Reimar Döffinger [Wed, 4 Mar 2009 18:18:49 +0000 (18:18 +0000)]
100l, interplayvideo uses 8x8 blocks, so call the 8x8 put_pixels_tab function
instead of the 16x16 one. Seems to fix decoding on PPC.

Originally committed as revision 17808 to svn://

13 years agoSet AVFrame.reference correctly for nuv decoder.
Reimar Döffinger [Wed, 4 Mar 2009 17:12:58 +0000 (17:12 +0000)]
Set AVFrame.reference correctly for nuv decoder.

Originally committed as revision 17807 to svn://

13 years agoAdd more correct bound checking in filter_mb().
Vitor Sessak [Wed, 4 Mar 2009 16:56:55 +0000 (16:56 +0000)]
Add more correct bound checking in filter_mb().

Should fix decoding of CVMAPAQP3_Sony_E.jsv in MinGW

Originally committed as revision 17806 to svn://

13 years agoAdd frame buffer allocators. aka simplify calls to
Gwenole Beauchesne [Wed, 4 Mar 2009 08:47:29 +0000 (08:47 +0000)]
Add frame buffer allocators. aka simplify calls to

Originally committed as revision 17804 to svn://

13 years agoMOV: fix demuxing fragmented files
Alex Converse [Wed, 4 Mar 2009 07:57:09 +0000 (07:57 +0000)]
MOV: fix demuxing fragmented files

Originally committed as revision 17803 to svn://

13 years agoFix a condition that causes an infinite loop, but only when compiled with
Mike Melanson [Wed, 4 Mar 2009 05:40:53 +0000 (05:40 +0000)]
Fix a condition that causes an infinite loop, but only when compiled with
gcc 4.2.4 on x86_32. Thanks to Vitor for hunting this down.

Originally committed as revision 17802 to svn://

13 years agoexcellent first pass at a description; now it's time for the Ministry of
Mike Melanson [Wed, 4 Mar 2009 05:24:59 +0000 (05:24 +0000)]
excellent first pass at a description; now it's time for the Ministry of
English Composition to tear it apart and rebuild it, stronger than before

Originally committed as revision 17801 to svn://

13 years agoPart 2 of 2 of Kenan Gillet's 'make ff_qcelp_lspf2lpc
Reynaldo H. Verdejo Pinochet [Wed, 4 Mar 2009 01:33:56 +0000 (01:33 +0000)]
Part 2 of 2 of Kenan Gillet's 'make ff_qcelp_lspf2lpc
more general' changeset. This one relocates the QCELP
especific code to the qcelpdec.* files.

Originally committed as revision 17800 to svn://

13 years agoPart 1 of 2 of Kenan Gillet's 'make ff_qcelp_lspf2lpc
Reynaldo H. Verdejo Pinochet [Wed, 4 Mar 2009 01:28:07 +0000 (01:28 +0000)]
Part 1 of 2 of Kenan Gillet's 'make ff_qcelp_lspf2lpc
more general' changeset. This one splits ff_qcelp_lspf2lpc
into the QCELP-especific ff_qcelp_lspf2lpc + the more general

Originally committed as revision 17799 to svn://

13 years agoflacdec: When there is not enough data to read the next header block,
Justin Ruggles [Wed, 4 Mar 2009 01:08:26 +0000 (01:08 +0000)]
flacdec: When there is not enough data to read the next header block,
reset the bitstream reader instead of skipping backwards.

Originally committed as revision 17798 to svn://

13 years agoSplit rtsp_send_cmd() into two functions, one for the actual sending of the
Ronald S. Bultje [Wed, 4 Mar 2009 00:57:37 +0000 (00:57 +0000)]
Split rtsp_send_cmd() into two functions, one for the actual sending of the
command and a second, new function to read the reply to this command. This
will make it possible to read server notices that are not in response to a
command in future versions, such as EOS or interrupt notices. See "[PATCH]
rtsp.c: split rtsp_send_cmd() in a send- and a receive-function" thread.

Originally committed as revision 17797 to svn://

13 years agocosmetics: indentation after last commit
Justin Ruggles [Wed, 4 Mar 2009 00:53:04 +0000 (00:53 +0000)]
cosmetics: indentation after last commit

Originally committed as revision 17796 to svn://

13 years agoflacdec: Check for an inline header before calling metadata_parse().
Justin Ruggles [Wed, 4 Mar 2009 00:52:18 +0000 (00:52 +0000)]
flacdec: Check for an inline header before calling metadata_parse().

Originally committed as revision 17795 to svn://

13 years agoflacdec: Allocate buffers right after STREAMINFO is parsed.
Justin Ruggles [Wed, 4 Mar 2009 00:21:46 +0000 (00:21 +0000)]
flacdec: Allocate buffers right after STREAMINFO is parsed.

Originally committed as revision 17794 to svn://

13 years agoAlpha: use -g3 with ccc to allow optimisation with debugging symbols
Måns Rullgård [Tue, 3 Mar 2009 23:53:29 +0000 (23:53 +0000)]
Alpha: use -g3 with ccc to allow optimisation with debugging symbols

Originally committed as revision 17793 to svn://

13 years agoFix syntax for the libfaad_decoder definition.
Stefano Sabatini [Tue, 3 Mar 2009 23:52:31 +0000 (23:52 +0000)]
Fix syntax for the libfaad_decoder definition.

Fix compilation (when libfaad is enabled).

Originally committed as revision 17792 to svn://

13 years agocosmetics: indentation after last commit
Justin Ruggles [Tue, 3 Mar 2009 23:50:21 +0000 (23:50 +0000)]
cosmetics: indentation after last commit

Originally committed as revision 17791 to svn://

13 years agoflacdec: Avoid parsing the STREAMINFO multiple times.
Justin Ruggles [Tue, 3 Mar 2009 23:49:36 +0000 (23:49 +0000)]
flacdec: Avoid parsing the STREAMINFO multiple times.

Originally committed as revision 17790 to svn://

13 years agoRename avctx_opts to avcodec_opts.
Stefano Sabatini [Tue, 3 Mar 2009 23:32:10 +0000 (23:32 +0000)]
Rename avctx_opts to avcodec_opts.

The new name is more meaningful and consistent with avformat_opts
and sws_opts.

Originally committed as revision 17789 to svn://

13 years agoAllow --enable-swscale until FATE is updated
Måns Rullgård [Tue, 3 Mar 2009 23:07:12 +0000 (23:07 +0000)]
Allow --enable-swscale until FATE is updated

Originally committed as revision 17788 to svn://

13 years agoRemove old scaler.
Michael Niedermayer [Tue, 3 Mar 2009 22:51:30 +0000 (22:51 +0000)]
Remove old scaler.

Originally committed as revision 17786 to svn://

13 years agoEnable AVFMT_FLAG_NONBLOCK.
Michael Niedermayer [Tue, 3 Mar 2009 20:25:21 +0000 (20:25 +0000)]

Originally committed as revision 17785 to svn://

13 years agoDo usleep(10000) when all demuxers returned EAGAIN.
Michael Niedermayer [Tue, 3 Mar 2009 20:24:58 +0000 (20:24 +0000)]
Do usleep(10000) when all demuxers returned EAGAIN.

Originally committed as revision 17784 to svn://

13 years agoReplace two 'return AVERROR(EAGAIN);' by continue. The latter are nicer
Michael Niedermayer [Tue, 3 Mar 2009 20:24:17 +0000 (20:24 +0000)]
Replace two 'return AVERROR(EAGAIN);' by continue. The latter are nicer
and the EAGAIN were just added as example not because they made sense.

Originally committed as revision 17783 to svn://

13 years agoIf a demuxer returns EAGAIN, try another demuxer.
Michael Niedermayer [Tue, 3 Mar 2009 19:50:04 +0000 (19:50 +0000)]
If a demuxer returns EAGAIN, try another demuxer.

Originally committed as revision 17782 to svn://

13 years agoremove the rest of x86 asm from LGPL build
Jindřich Makovička [Tue, 3 Mar 2009 19:47:23 +0000 (19:47 +0000)]
remove the rest of x86 asm from LGPL build

Originally committed as revision 28804 to svn://

13 years agoMOV: fix crash when 'meta' occurs before first 'trak'
Alex Converse [Tue, 3 Mar 2009 18:19:44 +0000 (18:19 +0000)]
MOV: fix crash when 'meta' occurs before first 'trak'

Originally committed as revision 17781 to svn://

13 years agoRename "fd1" variable ro "fd". There were previously two variables (fd1 and
Ronald S. Bultje [Tue, 3 Mar 2009 17:06:18 +0000 (17:06 +0000)]
Rename "fd1" variable ro "fd". There were previously two variables (fd1 and
fd2) and one was just removed, so naming the other "fd1" is counter-intuitive.
See "[RFC] rtsp.c EOF support" thread.

Originally committed as revision 17780 to svn://

13 years agoAdd url_get_file_handle(), which is used to get the file descriptor
Ronald S. Bultje [Tue, 3 Mar 2009 17:04:51 +0000 (17:04 +0000)]
Add url_get_file_handle(), which is used to get the file descriptor
associated with the I/O handle (e.g. the fd returned by open()). See
"[RFC] rtsp.c EOF support" thread.

There were previously some URI-specific implementations of the same idea,
e.g. rtp_get_file_handles() and udp_get_file_handle(). All of these are
deprecated by this patch and will be removed at the next major API bump.

Originally committed as revision 17779 to svn://

13 years agoReindent after r17777.
Ronald S. Bultje [Tue, 3 Mar 2009 16:53:04 +0000 (16:53 +0000)]
Reindent after r17777.

Originally committed as revision 17778 to svn://

13 years agoMake RTSP-MS-over-UDP negotiation work. See "[PATCH] RTSP-MS 8/15: fix
Ronald S. Bultje [Tue, 3 Mar 2009 16:52:35 +0000 (16:52 +0000)]
Make RTSP-MS-over-UDP negotiation work. See "[PATCH] RTSP-MS 8/15: fix
RTSP-MS UDP" thread on mailinglist.

Basically, UDP setup needs to be done in a particular order (first rtx
on two UDP ports (one for RTP, one for RTCP), then the other streams over
one, single port for all of them together). Not doing this correctly results
in a "461" error (invalid transport) during setup.

Originally committed as revision 17777 to svn://

13 years agoRecognize the "application" data type, which is required for WMS/UDP
Ronald S. Bultje [Tue, 3 Mar 2009 16:48:56 +0000 (16:48 +0000)]
Recognize the "application" data type, which is required for WMS/UDP

This type is used in RTP/ASF (served by WMS servers), and is required to
make UDP sessions work, but breaks TCP sessions. Therefore, we disable setup
for application streams in TCP/WMS streams.

See discussion in "[PATCH] RTSP-MS 8/15: fix RTSP-MS UDP" thread.

Originally committed as revision 17776 to svn://

13 years agoAdd a context to av_log() calls.
Benoit Fouet [Tue, 3 Mar 2009 16:19:37 +0000 (16:19 +0000)]
Add a context to av_log() calls.

Originally committed as revision 17775 to svn://

13 years agoA quick description of Rate distortion theory.
Michael Niedermayer [Tue, 3 Mar 2009 16:09:19 +0000 (16:09 +0000)]
A quick description of Rate distortion theory.

Originally committed as revision 17774 to svn://

13 years agoDo not use a macro to generate a single AVCodec declaration.
Diego Biurrun [Tue, 3 Mar 2009 15:48:52 +0000 (15:48 +0000)]
Do not use a macro to generate a single AVCodec declaration.

Originally committed as revision 17773 to svn://

13 years agoQuick description of the Viterbi algorithm so I do not have to repeat it
Michael Niedermayer [Tue, 3 Mar 2009 15:35:20 +0000 (15:35 +0000)]
Quick description of the Viterbi algorithm so I do not have to repeat it
over and over again on the ML.

Originally committed as revision 17772 to svn://

13 years agoWhen a H.264 stream references a PPS or SPS id which doesn't exist, instead of
Panagiotis Issaris [Tue, 3 Mar 2009 14:32:24 +0000 (14:32 +0000)]
When a H.264 stream references a PPS or SPS id which doesn't exist, instead of
just saying that a non-existing id is referenced, show the value of the id.

Originally committed as revision 17771 to svn://

13 years agoAdd vhook removal to the changelog.
Diego Biurrun [Tue, 3 Mar 2009 14:12:41 +0000 (14:12 +0000)]
Add vhook removal to the changelog.

Originally committed as revision 17770 to svn://

13 years agoRemove deprecated vhook subsystem.
Diego Biurrun [Tue, 3 Mar 2009 14:09:10 +0000 (14:09 +0000)]
Remove deprecated vhook subsystem.

Originally committed as revision 17769 to svn://

13 years agoRemove size_t cast in setting s->priv_data directly to the (integer) file
Ronald S. Bultje [Tue, 3 Mar 2009 13:57:09 +0000 (13:57 +0000)]
Remove size_t cast in setting s->priv_data directly to the (integer) file
descriptor returned by open(). This removes some dubious doublecasts such
as priv_data = (void *) (size_t) some_integer, and is always safe on systems
we care about because sizeof(int)<=sizeof(void*). See comments from Mans and
Michael in "[RFC] rtsp.c EOF support" thread.

Originally committed as revision 17768 to svn://

13 years agoDon't let finalize_packet() touch pkt->stream_index. Instead, let individual
Ronald S. Bultje [Tue, 3 Mar 2009 13:51:34 +0000 (13:51 +0000)]
Don't let finalize_packet() touch pkt->stream_index. Instead, let individual
payload handlers take care of that themselves at their own option. What this
patch really does is "fix" a bug in MS-RTSP protocol where incoming packets
are always coming in over the connection (UDP) or interleave-id (TCP) of
the stream-id of the first ASF packet in the RTP packet. However, RTP packets
may contain multiple ASF packets (and usually do, from what I can see), and
therefore this leads to playback bugs. The intended stream-id per ASF packet
is given in the respective ASF packet header. The ASF demuxer will correctly
read this and set pkt->stream_index, but since the "stream" parameter can
not be known to rtpdec.c or any of the RTP/RTSP code, the "st" parameter
in all these functions is basically invalid. Therefore, using st->id as
pkt->stream_index leads to various playback bugs. The result of this patch
is that pkt->stream_index is left untouched for RTP/ASF (and possibly for
other payloads that have similar behaviour).

The patch was discussed in the "[PATCH] rtpdec.c: don't overwrite
pkt->stream_index in finalize_packet()" thread on the mailinglist.

Originally committed as revision 17767 to svn://

13 years agoSet PixFmtInfo::{x, y}_chroma_shift for VDPAU and VAAPI formats.
Gwenole Beauchesne [Tue, 3 Mar 2009 13:43:24 +0000 (13:43 +0000)]
Set PixFmtInfo::{x, y}_chroma_shift for VDPAU and VAAPI formats.

Originally committed as revision 17766 to svn://

13 years agoReindent after r17764.
Ronald S. Bultje [Tue, 3 Mar 2009 13:42:16 +0000 (13:42 +0000)]
Reindent after r17764.

Originally committed as revision 17765 to svn://

13 years agoIn the current implementation of rtp_parse_packet(), finalize_packet() is
Ronald S. Bultje [Tue, 3 Mar 2009 13:41:50 +0000 (13:41 +0000)]
In the current implementation of rtp_parse_packet(), finalize_packet() is
called for all packets with an internal handler function but only for
non-first packets from dynamic payload parse_packet() handlers. This patch
fixes that. Bug was noticed by Luca in "[PATCH] rtpdec.c: don't overwrite
pkt->stream_index in finalize_packet()" thread.

Originally committed as revision 17764 to svn://

13 years agoStart Changelog for next version.
Diego Biurrun [Tue, 3 Mar 2009 13:38:40 +0000 (13:38 +0000)]
Start Changelog for next version.

Originally committed as revision 17763 to svn://

13 years agocosmetics: Add some empty lines for readability.
Diego Biurrun [Tue, 3 Mar 2009 13:30:57 +0000 (13:30 +0000)]
cosmetics: Add some empty lines for readability.

Originally committed as revision 17762 to svn://

13 years agoReduce allocated length of the HTTP authentication request field buffer, as
Ronald S. Bultje [Tue, 3 Mar 2009 13:26:17 +0000 (13:26 +0000)]
Reduce allocated length of the HTTP authentication request field buffer, as
noticed by Stefano and Luca in the "[PATCH]RTSP Basic Authentication"
mailinglist thread.

av_base64_encode() was recently changed. The previous implementation required
12 extra bytes (ceil(len(src)/3.)*4+12), whereas the new one is guaranteed to
fit in an exact buffer (ceil(len(src)/3.)*4), plus one extra byte for the
trailing zero. This change fixes no bug, it just slightly decreases the
amount of allocated memory.

Originally committed as revision 17761 to svn://

13 years agoCosmetics: Typo
Panagiotis Issaris [Tue, 3 Mar 2009 13:25:53 +0000 (13:25 +0000)]
Cosmetics: Typo

Originally committed as revision 17760 to svn://

13 years agoUse sign_extend().
Benoit Fouet [Tue, 3 Mar 2009 13:11:10 +0000 (13:11 +0000)]
Use sign_extend().
On behalf of Jai.

Originally committed as revision 17759 to svn://

13 years agoAdd my GPG fingerprint
Panagiotis Issaris [Tue, 3 Mar 2009 13:01:38 +0000 (13:01 +0000)]
Add my GPG fingerprint

Originally committed as revision 17758 to svn://

13 years agoCosmetics: Typo
Panagiotis Issaris [Tue, 3 Mar 2009 12:58:49 +0000 (12:58 +0000)]
Cosmetics: Typo

Originally committed as revision 17757 to svn://

13 years agocosmetics: Fix indentation.
Diego Biurrun [Tue, 3 Mar 2009 12:57:07 +0000 (12:57 +0000)]
cosmetics: Fix indentation.

Originally committed as revision 17756 to svn://

13 years agoDo not add -D_FILE_OFFSET_BITS=64 and -D_LARGEFILE_SOURCE unconditionally to
Diego Biurrun [Tue, 3 Mar 2009 12:52:12 +0000 (12:52 +0000)]
Do not add -D_FILE_OFFSET_BITS=64 and -D_LARGEFILE_SOURCE unconditionally to
CFLAGS.  Apparently there are some systems that do not like these definitions.

Originally committed as revision 17755 to svn://

13 years agoRemove mpeg4aac dependency declaration, the decoder was removed.
Diego Biurrun [Tue, 3 Mar 2009 12:45:48 +0000 (12:45 +0000)]
Remove mpeg4aac dependency declaration, the decoder was removed.

Originally committed as revision 17753 to svn://

13 years agoconfigure: improve temp file creation and cleanup
Måns Rullgård [Tue, 3 Mar 2009 08:53:55 +0000 (08:53 +0000)]
configure: improve temp file creation and cleanup

Originally committed as revision 17752 to svn://

13 years agoflacdec: Warn about invalid max blocksize and limit the minimum value.
Justin Ruggles [Tue, 3 Mar 2009 05:53:38 +0000 (05:53 +0000)]
flacdec: Warn about invalid max blocksize and limit the minimum value.

Originally committed as revision 17751 to svn://

13 years agoH.264: Simplify decode_residual()
Alex Converse [Tue, 3 Mar 2009 05:26:39 +0000 (05:26 +0000)]
H.264: Simplify decode_residual()

Originally committed as revision 17750 to svn://

13 years agocosmetics: remove a pointless comment
Justin Ruggles [Tue, 3 Mar 2009 05:25:58 +0000 (05:25 +0000)]
cosmetics: remove a pointless comment

Originally committed as revision 17749 to svn://

13 years agoflacdec: Remove unused variable, min_blocksize.
Justin Ruggles [Tue, 3 Mar 2009 05:25:23 +0000 (05:25 +0000)]
flacdec: Remove unused variable, min_blocksize.

Originally committed as revision 17748 to svn://

13 years agoflacdec: Return error when blocksize code of 0 is found. It is a
Justin Ruggles [Tue, 3 Mar 2009 05:22:44 +0000 (05:22 +0000)]
flacdec: Return error when blocksize code of 0 is found. It is a
reserved value per the FLAC format documentation.

Originally committed as revision 17747 to svn://

13 years agoEnable PB-frames decoding for H.263
Kostya Shishkov [Tue, 3 Mar 2009 05:18:39 +0000 (05:18 +0000)]
Enable PB-frames decoding for H.263

Originally committed as revision 17746 to svn://

13 years agoAdd PB-frame decoding support for H.263/i263 (and enable it for i263)
Kostya Shishkov [Tue, 3 Mar 2009 05:17:22 +0000 (05:17 +0000)]
Add PB-frame decoding support for H.263/i263 (and enable it for i263)

Originally committed as revision 17745 to svn://

13 years agoalacdec: Simplify reading of uncompressed samples by using
Justin Ruggles [Tue, 3 Mar 2009 04:57:02 +0000 (04:57 +0000)]
alacdec: Simplify reading of uncompressed samples by using

Originally committed as revision 17744 to svn://

13 years agoflacdec: Avoid trying to read 33 bits. This occurs when the source
Justin Ruggles [Tue, 3 Mar 2009 04:45:45 +0000 (04:45 +0000)]
flacdec: Avoid trying to read 33 bits.  This occurs when the source
is 32 bits-per-sample and channel decorrelation is used.  Such files
are valid, but not supported currently.

Originally committed as revision 17743 to svn://

13 years agoflacdec: ALT_BITSTREAM_READER is no longer required.
Justin Ruggles [Tue, 3 Mar 2009 02:53:11 +0000 (02:53 +0000)]
flacdec: ALT_BITSTREAM_READER is no longer required.

Originally committed as revision 17742 to svn://

13 years agoflacdec: use get_sbits_long() where needed.
Justin Ruggles [Tue, 3 Mar 2009 02:19:01 +0000 (02:19 +0000)]
flacdec: use get_sbits_long() where needed.

Originally committed as revision 17741 to svn://

13 years agoAdd a get_sbits_long() function.
Justin Ruggles [Tue, 3 Mar 2009 02:00:47 +0000 (02:00 +0000)]
Add a get_sbits_long() function.

Originally committed as revision 17740 to svn://

13 years agoALAC: use sign_extend() from mathops.h
Måns Rullgård [Tue, 3 Mar 2009 01:53:54 +0000 (01:53 +0000)]
ALAC: use sign_extend() from mathops.h

Originally committed as revision 17739 to svn://

13 years agoAdd sign_extend() function to mathops.h
Måns Rullgård [Tue, 3 Mar 2009 00:03:55 +0000 (00:03 +0000)]
Add sign_extend() function to mathops.h

Originally committed as revision 17738 to svn://

13 years agoSimplify show_bits_long and copy the GetBitsContext around only once instead of twice.
Reimar Döffinger [Mon, 2 Mar 2009 19:42:31 +0000 (19:42 +0000)]
Simplify show_bits_long and copy the GetBitsContext around only once instead of twice.

Originally committed as revision 17736 to svn://

13 years agoflacdec: Use get/skip_bits_long() for more than 17-bits and
Justin Ruggles [Mon, 2 Mar 2009 17:32:53 +0000 (17:32 +0000)]
flacdec: Use get/skip_bits_long() for more than 17-bits and
get/skip_bits() for 17-bits or less.

Originally committed as revision 17735 to svn://

13 years agoFactor out block decoding in ff_h263_decode_mb()
Kostya Shishkov [Mon, 2 Mar 2009 17:18:49 +0000 (17:18 +0000)]
Factor out block decoding in ff_h263_decode_mb()

Originally committed as revision 17734 to svn://

13 years agoConsider all packets in the parser, not just ones with timestamps,
Michael Niedermayer [Mon, 2 Mar 2009 15:13:58 +0000 (15:13 +0000)]
Consider all packets in the parser, not just ones with timestamps,
this should greatly simplify handling of packet pos.

Originally committed as revision 17733 to svn://

13 years agoFavor container packets that end after the first byte of the access
Michael Niedermayer [Mon, 2 Mar 2009 15:12:26 +0000 (15:12 +0000)]
Favor container packets that end after the first byte of the access
unit in fetch_timestamp(). This should make no difference for valid
streams but may help invalid ones, also its needed for future changes.

Originally committed as revision 17732 to svn://

13 years agoCall ff_fetch_timestamp() for mpeg1/2 when a picture start code is found instead
Michael Niedermayer [Mon, 2 Mar 2009 14:53:18 +0000 (14:53 +0000)]
Call ff_fetch_timestamp() for mpeg1/2 when a picture start code is found instead
of calling it at the end of a frame with a large negative offset.
This significantly reduces the maximal distance in container packets between
the point where the first byte of the "access unit" was stored and where
we call ff_fetch_timestamp() thus reducing the constraints on our parser.
Also change the parser from next_frame_offset to cur, this is needed
because now the reference is from container packet start instead of
frame start. (i previously misinterpreted this as bug)

Originally committed as revision 17731 to svn://

13 years agoversion.h should depend on config.mak.
Diego Biurrun [Mon, 2 Mar 2009 09:33:57 +0000 (09:33 +0000)]
version.h should depend on config.mak.
The extra version string that is passed to configure is recorded in config.mak
and changes to config.mak can therefore make version.h rebuilds necessary.

Originally committed as revision 17730 to svn://

13 years agoAdd context to some av_log() calls.
Benoit Fouet [Mon, 2 Mar 2009 09:28:37 +0000 (09:28 +0000)]
Add context to some av_log() calls.

Originally committed as revision 17729 to svn://

13 years agoMerge three conditions in a single 'if' instead of two.
Benoit Fouet [Mon, 2 Mar 2009 09:22:17 +0000 (09:22 +0000)]
Merge three conditions in a single 'if' instead of two.

Originally committed as revision 17728 to svn://

13 years agoMention non-recursive Makefiles in the changelog.
Diego Biurrun [Mon, 2 Mar 2009 07:57:04 +0000 (07:57 +0000)]
Mention non-recursive Makefiles in the changelog.

Originally committed as revision 17725 to svn://

13 years agocosmetics: spelling/wording fixes
Diego Biurrun [Mon, 2 Mar 2009 07:52:47 +0000 (07:52 +0000)]
cosmetics: spelling/wording fixes

Originally committed as revision 17724 to svn://

13 years agoAC-3 encoding is integer-only
Justin Ruggles [Mon, 2 Mar 2009 06:31:28 +0000 (06:31 +0000)]
AC-3 encoding is integer-only

Originally committed as revision 17723 to svn://

13 years agoRevert previous removal of PAM, PBM, PGM, PGMYUV and PPM from the list of
Diego Biurrun [Mon, 2 Mar 2009 06:13:54 +0000 (06:13 +0000)]
Revert previous removal of PAM, PBM, PGM, PGMYUV and PPM from the list of
supported image formats. Apparently decoding these formats is supported.

Originally committed as revision 17722 to svn://

13 years agocosmetics: Reformat long_names so that "DPCM" comes first.
Diego Biurrun [Mon, 2 Mar 2009 06:01:32 +0000 (06:01 +0000)]
cosmetics: Reformat long_names so that "DPCM" comes first.

Originally committed as revision 17721 to svn://

13 years agocosmetics: Reformat long_names so that "ADPCM" comes first.
Diego Biurrun [Mon, 2 Mar 2009 05:55:03 +0000 (05:55 +0000)]
cosmetics: Reformat long_names so that "ADPCM" comes first.

Originally committed as revision 17720 to svn://

13 years agocosmetics: Reformat long_names so that "PCM" comes first.
Diego Biurrun [Mon, 2 Mar 2009 05:45:44 +0000 (05:45 +0000)]
cosmetics: Reformat long_names so that "PCM" comes first.

Originally committed as revision 17719 to svn://