[Ffmpeg-devel-irc] ffmpeg-devel.log.20120505

Sun May 6 02:05:04 CEST 2012

[00:01] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * r1f95ad48ff 10ffmpeg/ (libavcodec/vorbis_parser.c libavformat/oggparsevorbis.c): 
[00:01] <CIA-122> ffmpeg: oggvorbisdec: redesign special handling of first frames timestamp.
[00:01] <CIA-122> ffmpeg: This also will make it possible to use the code for cases other than the start of a stream
[00:01] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:01] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * rfe5c5bccce 10ffmpeg/libavformat/oggparsevorbis.c: 
[00:01] <CIA-122> ffmpeg: oggvorbisdec: Apply timestamp calculation always when timestamps arent known.
[00:01] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:01] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * rb18c9f1eb0 10ffmpeg/libavformat/oggparsetheora.c: 
[00:01] <CIA-122> ffmpeg: oggtheora: Port changes from oggvorbis timestamp handling.
[00:01] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:58] <CIA-122> ffmpeg: 03Mans Rullgard 07master * rc02efacc8f 10ffmpeg/libavutil/arm/intreadwrite.h: 
[00:58] <CIA-122> ffmpeg: arm: intreadwrite: revert 16-bit load asm to old version for gcc < 4.6
[00:58] <CIA-122> ffmpeg: Commit adebad0 "arm: intreadwrite: fix inline asm constraints for gcc
[00:58] <CIA-122> ffmpeg: 4.6 and later" caused some older gcc versions to miscompile code.
[00:58] <CIA-122> ffmpeg: This reverts to the old version of the code for these compilers.
[00:58] <CIA-122> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[00:58] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * rbce42e95d1 10ffmpeg/: (log message trimmed)
[00:58] <CIA-122> ffmpeg: Merge remote-tracking branch 'qatar/master'
[00:58] <CIA-122> ffmpeg: * qatar/master:
[00:58] <CIA-122> ffmpeg:  configure: add POWER[5-7] support
[00:58] <CIA-122> ffmpeg:  arm: intreadwrite: revert 16-bit load asm to old version for gcc < 4.6
[00:58] <CIA-122> ffmpeg:  vqavideo: return error if image size is not a multiple of block size
[00:58] <CIA-122> ffmpeg:  cosmetics: indentation
[00:58] <CIA-122> ffmpeg: 03Sean McGovern 07master * r313f9fbfbb 10ffmpeg/configure: 
[00:58] <CIA-122> ffmpeg: configure: add POWER[5-7] support
[00:58] <CIA-122> ffmpeg: Also merge POWER3 and POWER4 configuration together with the additions.
[00:58] <CIA-122> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[00:58] <CIA-122> ffmpeg: This modifies pts in situations other than what was intended, leading to
[00:58] <CIA-122> ffmpeg: invalid timestamps.
[00:58] <CIA-122> ffmpeg: Reverts commit 90bb394dccacd10607153833a0aeba0d970dc8db
[00:58] <CIA-122> ffmpeg: 03Justin Ruggles 07master * r8916f1fbcb 10ffmpeg/libavformat/utils.c: 
[00:58] <CIA-122> ffmpeg: avformat: only fill-in interpolated timestamps if duration is non-zero
[00:58] <CIA-122> ffmpeg: This avoids returning duplicate timestamps for multiple packets when the
[00:58] <CIA-122> ffmpeg: demuxer does not provide all timestamps and packet duration is not known.
[02:07] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * r441dce2169 10ffmpeg/ (libavcodec/vorbis_parser.c libavformat/oggparsevorbis.c): 
[02:07] <CIA-122> ffmpeg: oggvorbis: move handling of first packets ts from parser to muxer.
[02:07] <CIA-122> ffmpeg: The parser does not have enough knowledge it seems to do it
[02:07] <CIA-122> ffmpeg: correctly.
[02:07] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:07] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * r16db88a093 10ffmpeg/libavcodec/vorbis_parser.c: 
[02:07] <CIA-122> ffmpeg: vorbis_parser: fix blocksize
[02:07] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:16] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * r3d42addf33 10ffmpeg/libavcodec/png_parser.c: 
[02:16] <CIA-122> ffmpeg: png_parser: try to fix big endian
[02:16] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[11:12] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * r45a7b0674d 10ffmpeg/libavformat/oggparsevorbis.c: 
[11:12] <CIA-122> ffmpeg: oggvorbis: fix the first 2 packets timestamps matching issue
[11:12] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[11:43] <cbsrobot_> after reading http://multimedia.cx/eggs/yuv4mpeg2-origin/ is there any reason yuv4mpeg should not support yuv pix_fmts > 8 ? 
[11:48] <JEEB> is there a bit depth flag?
[11:48] <cbsrobot_> the header contains the pix_fmt
[13:24] <CIA-122> ffmpeg: 03Carl Eugen Hoyos 07master * r849172882c 10ffmpeg/libavcodec/alsdec.c: 
[13:24] <CIA-122> ffmpeg: alsdec.c: Simplify error check before calling rice decoder.
[13:24] <CIA-122> ffmpeg: Reviewed-by: Thilo Borgmann
[13:24] <CIA-122> ffmpeg: 03Carl Eugen Hoyos 07master * r396d2aecc5 10ffmpeg/libavcodec/alsdec.c: 
[13:24] <CIA-122> ffmpeg: alsdec.c: Also test 0 when checking rice decoder values.
[13:24] <CIA-122> ffmpeg: Suggested-by: Thilo Borgmann
[14:50] <CIA-122> ffmpeg: 03Carl Eugen Hoyos 07master * rc6758ac027 10ffmpeg/libavformat/mpc8.c: 
[14:50] <CIA-122> ffmpeg: Skip padding bytes after reading musepack8 header.
[14:50] <CIA-122> ffmpeg: Fixes ticket #1160.
[14:55] <ubitux> question: let's say there are two different subtitles formats, one with the following syntax at each line: "[start=hh:mm:ss][end=hh:mm:ss] text" and the other one: "#hh:mm:ss#hh:mm:ss# text"
[14:55] <CIA-122> ffmpeg: 03Carl Eugen Hoyos 07master * r23fba3ed5c 10ffmpeg/libavcodec/ffv1.c: Remove unused variable from ffv1 decoder.
[14:55] <ubitux> would it make sense for the two demuxer to raise a packet with only "text" in it
[14:55] <ubitux> and a common generic subtitles decoder?
[14:56] <ubitux> (the packet already has start and duration information, so the muxer can reconstruct them)
[14:56] <ubitux> i'm wondering how this will affect the "pos" attribute and the heuristic around this when seeking
[18:43] <Daemon404> wtf
[18:43] <Daemon404> 444 new emails in ffmpeg-devel
[18:43] <nevcairiel> you must be broken, i only see 1
[18:44] <Daemon404> i do see some chinese spam
[18:44] <Daemon404> i think
[18:44] <nevcairiel> i had 4 mails of those or so
[18:44] <nevcairiel> maybe gmail filtered the others automatically?
[18:45] <nevcairiel> ubitux: isnt that how pure text subs in mkv for example already work? just pure text in the packet? (re subtitles earlier)
[18:46] <Daemon404> btw 444 is from teh last 12 hrs
[18:46] <nevcairiel> hard to count, but certainly under 30 :P
[18:47] <Daemon404> lol no
[18:47] <nevcairiel> 400 of those spam? :)
[18:47] <Daemon404> there are a lot of large email threads :P
[18:47] <Daemon404> 12+ hrs like i said
[18:48] <nevcairiel> either i am not getting alot of the mails, or i dont know
[18:48] <burek> if you use gmail box, check spam folder
[18:48] Action: nevcairiel checks the web archive
[18:48] <burek> they were testing new filters last few months
[18:49] <burek> so those mails might have gotten stuck
[18:49] <Daemon404> im using gmail too
[18:49] <Daemon404> but not web
[18:49] <burek> imap?
[18:49] <nevcairiel> the web archive of the mailing list mentions 244 mails this month
[18:49] <nevcairiel> certainly not 444 in the last day
[18:49] <nevcairiel> :P
[18:50] <Daemon404> lol maybe my client derped
[18:50] <Daemon404> and marked some as unread
[18:50] <nevcairiel> possibly
[18:53] <burek> is it true that h265 is coming out soon or is it just a rumor?
[18:54] <nevcairiel> the spec is scheduled to be finalized in january 2013, which then still needs to be ratified as a standard
[18:54] <burek> I've heard something about doubling the compression with the same image quality and such stuff...
[18:54] <burek> man that will be awesome
[18:54] <nevcairiel> thats their goal
[18:55] <Daemon404> keep in mind
[18:55] <Daemon404> its still a good amount of the usual pimping of things
[19:07] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * r47055b8913 10ffmpeg/libswresample/x86/ (audio_convert.asm swresample_x86.c): 
[19:07] <CIA-122> ffmpeg: swr: implement stereo S16/S32/FLT->S16/S32/FLT planar->packed in SSE/SSE2
[19:07] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:07] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * re8dd7928c8 10ffmpeg/libswresample/ (audioconvert.c x86/audio_convert.asm): 
[19:07] <CIA-122> ffmpeg: swr: change simd len argument to be in samples instead of dst bytes.
[19:07] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:07] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * rfec3700dcd 10ffmpeg/libswresample/audioconvert.c: 
[19:07] <CIA-122> ffmpeg: swr: update calling code to support mixed packed planar SIMD
[19:07] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:07] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * r3db6093244 10ffmpeg/libswresample/swresample_test.c: 
[19:07] <CIA-122> ffmpeg: swr-test: allow testing a specific subset of cases
[19:07] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:32] <ubitux> nevcairiel: mmh that's possible, i was just wondering if it would make sense to cut off a lot of data while demuxing
[19:33] <ubitux> and also, if the pos attribute should point on the beginning of the text, or the "full" packet
[19:47] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * raae3119643 10ffmpeg/libswresample/x86/audio_convert.asm: 
[19:47] <CIA-122> ffmpeg: swr: replace planar->planar/packed->packed FLT<->S16/S32 SIMD by new macros
[19:47] <CIA-122> ffmpeg: this simplifies the code
[19:47] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:47] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * rb6f4f0d9ef 10ffmpeg/libswresample/x86/audio_convert.asm: 
[19:47] <CIA-122> ffmpeg: swr: fix PACK_2CH register count
[19:47] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:47] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * rdf5ff103cd 10ffmpeg/libswresample/x86/audio_convert.asm: 
[19:48] <CIA-122> ffmpeg: swr: fix internal asm labels
[19:48] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:03] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * r5f4e18cd16 10ffmpeg/libswresample/x86/audio_convert.asm: 
[20:03] <CIA-122> ffmpeg: swr: replace the remaining 2 audio convert SIMD macros by the new ones
[20:03] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:36] <CIA-122> ffmpeg: 03Ronald S. Bultje 07master * rd2205d6543 10ffmpeg/libavcodec/pngdec.c: 
[20:36] <CIA-122> ffmpeg: png: check bit depth for PAL8/Y400A pixel formats.
[20:36] <CIA-122> ffmpeg: Wrong bit depth can lead to invalid rowsize values, which crashes the
[20:36] <CIA-122> ffmpeg: decoder further down.
[20:36] <CIA-122> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[20:36] <CIA-122> ffmpeg: CC: libav-stable at libav.org
[20:36] <CIA-122> ffmpeg: 03Ronald S. Bultje 07master * r273e6af47b 10ffmpeg/libavformat/electronicarts.c: 
[20:36] <CIA-122> ffmpeg: ea: check chunk_size for validity.
[20:36] <CIA-122> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[20:36] <CIA-122> ffmpeg: CC: libav-stable at libav.org
[20:36] <CIA-122> ffmpeg: 03Alex Converse 07master * r37ddd38332 10ffmpeg/libavcodec/celp_filters.c: 
[20:36] <CIA-122> ffmpeg: celp filters: Do not read earlier than the start of the 'out' vector.
[20:36] <CIA-122> ffmpeg: CC: libav-stable at libav.org
[20:36] <CIA-122> ffmpeg: 03Martin Storsjö 07master * r5245adb963 10ffmpeg/libavformat/rtpdec_h264.c: 
[20:36] <CIA-122> ffmpeg: rtpdec_h264: Check the available data length before reading
[20:36] <CIA-122> ffmpeg: This makes sure the length is checked for STAP-A type packets.
[20:37] <CIA-122> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[20:37] <CIA-122> ffmpeg: 03Ivan Kovtunov 07master * rde26a4b699 10ffmpeg/libavformat/rtpdec_h264.c: 
[20:37] <CIA-122> ffmpeg: rtpdec_h264: Add input size checks
[20:37] <CIA-122> ffmpeg: This fixes crashes if given too short data packets.
[20:37] <CIA-122> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[20:37] <CIA-122> ffmpeg: 03Jakub Stachowski 07master * rddffe3de43 10ffmpeg/libavcodec/wmalosslessdec.c: 
[20:37] <CIA-122> ffmpeg: WMAL: Shift output samples by the specified number of padding zeroes.
[20:37] <CIA-122> ffmpeg: Signed-off-by: Kostya Shishkov <kostya.shishkov at gmail.com>
[20:37] <CIA-122> ffmpeg: 03Mashiat Sarker Shakkhar 07master * r363c3a44ff 10ffmpeg/libavcodec/wmalosslessdec.c: 
[20:37] <CIA-122> ffmpeg: WMAL: Restore removed code in mclms_predict()
[20:37] <CIA-122> ffmpeg: Based on observations made by Jakub Stachowski <qbast at go2.pl>
[20:37] <CIA-122> ffmpeg: Signed-off-by: Kostya Shishkov <kostya.shishkov at gmail.com>
[20:37] <CIA-122> ffmpeg: 03Martin Storsjö 07master * r8d43b8b8e8 10ffmpeg/libavformat/rtpdec_h264.c: 
[20:37] <CIA-122> ffmpeg: rtpdec_h264: Remove outdated/useless/incorrect comments
[20:37] <CIA-122> ffmpeg: RTCP is handled elsewhere, not in the depacketizer for an
[20:37] <CIA-122> ffmpeg: individual format.
[20:37] <CIA-122> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[21:39] Action: ubitux just realizes that the insane crap SAMI format is, doesn't even support a "end" timestamp
[21:39] <ubitux> seriously, what were those guys thinking
[21:39] <ubitux> every CSS property must be supported
[21:39] <Compn> oh like end at 999999 time ?
[21:40] <ubitux> but timestamp end, no, no problem.
[21:40] <ubitux> Compn: end at the next sub even i guess
[21:42] <ubitux> maybe there is a trick with an empty sub
[21:42] <ubitux> but i can't find anything on http://msdn.microsoft.com/en-us/library/ms971327.aspx
[21:43] <Compn> just download all sami and look for workarounds in there ? :P
[22:02] <ubitux> mmh i wonder if this won't do the trick: To blank a paragraph, a non-breaking space entity ( ) is used.
[22:10] <burek> last words of a dying PC machine: "Starting Windows..." :)
[00:00] --- Sun May  6 2012