[Ffmpeg-devel-irc] ffmpeg-devel.log.20120423

[burek]

[07:43] <ubitux> http://ffmpeg.org/pipermail/ffmpeg-devel/2012-April/123616.html
[07:43] <ubitux> ’ http://lists.libav.org/pipermail/libav-devel/2012-April/026542.html
[07:44] <ubitux> funny to see again some ostrich policy
[07:44] <ubitux> and funny to see that even the reviews are the same
[07:46] <ubitux> i want my internet back to commit stuff :(
[08:01] <bryno> ubitux: interesting, indeed.
[08:05] <ohsix> ostritch policy?
[08:06] <ubitux> certainly a french expression; i meant "restraining from looking around" policy
[08:06] <ubitux> (ostritch with the head in the floor)
[08:06] <ohsix> i got that, but i'm not sure what you mean, that the typo was missed?
[08:07] <ubitux> i don't see credits to nicolas, or luca taking into account the typo reimar fixed at first
[08:08] <ohsix> ah
[08:08] <ubitux> the patch was ignored from ffmpeg again, and "re-invented"
[08:08] <ohsix> well they said they won't take anything from it, but they probably have someone looking to duplicate the work :]
[08:09] <ubitux> no, when it happens they at least keep the credits
[08:09] <ohsix> luca might have seen someone apply it to gentoo
[08:10] <ubitux> anyway, not the first time this stuff happens, it's too bad
[08:10] <ubitux> but well, too bad for them.
[09:53] <ubitux> a lot of mp3 overread in this sample, can anyone confirm http://kagerow.net/mp3/sakura.mp3 ?
[09:56] <ubitux> playback is fine though
[09:56] <ubitux> i guess it's been badly encoded
[10:29] <VooDooNOFX> When examining the magic headers of some recent r3d files, it appears they changed their magic number from DRED1 to fRED2. Does anyone know if this has ever been brought up?
[10:33] <TimNich> fRED2 or DRED2 ?
[10:34] <av500> Red Said Fred
[10:34] <VooDooNOFX> hexedit shows 00 00 01 66 52 45 44 32 05 04 52 32
[10:34] <VooDooNOFX> ...fRED2..R2
[10:35] <ubitux> lavf demuxer checks for "RED1"
[10:36] <ubitux> looks like they just bump the version to introduce new stuff
[10:36] <VooDooNOFX> yep, r3d_probe tests for RED1
[10:37] <ubitux> 00 00 01 66 is the size
[10:38] <ubitux> the fact it is 'D' or 'f' doesn't matter, just a coincidence
[10:40] <VooDooNOFX> is that a U32 size? or R32?
[10:41] <ubitux> u32 afaict
[10:41] <ubitux> (big-endian)
[10:41] <TimNich> I remember ARM's 'misuse' of hex to put 'DEADBEEF' to indicate a defunct file.....
[10:43] <VooDooNOFX> interesting. this file is around 3.1gb
[10:48] <ubitux> it doesn't look like the red1 fmt supports 64 bits atom sizes
[10:48] <ubitux> maybe the red2 added that
[10:52] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r4d87001096 10ffmpeg/libavcodec/vp8.c: 
[10:52] <CIA-17> ffmpeg: vp8: fix crash due to skiped update_dimensions().
[10:52] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[10:52] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[10:54] <VooDooNOFX> i'll update the magic string, recompile ffprobe and run it on the files i've got. Maybe it can still be read.
[13:16] <CIA-17> ffmpeg: 03Nicolas George 07master * r73bc019baf 10ffmpeg/libavformat/utils.c: lavf: print a warning if probesize seems not enough.
[14:30] <CIA-17> ffmpeg: 03Nicolas George 07master * r78ac49c233 10ffmpeg/libavfilter/avfilter.h: avfilter: document request_frame return codes.
[14:30] <CIA-17> ffmpeg: 03Nicolas George 07master * re0761feec4 10ffmpeg/libavfilter/ (avfilter.h avfiltergraph.c): avfilter: add a pointer from links to graph.
[14:30] <CIA-17> ffmpeg: 03Nicolas George 07master * r2d6522bab0 10ffmpeg/libavfilter/ (vsrc_buffer.c vsrc_buffer.h): vsrc_buffer: count the number of failed requests.
[14:30] <CIA-17> ffmpeg: 03Nicolas George 07master * re11110dee4 10ffmpeg/libavfilter/ (buffersink.h sink_buffer.c version.h): buffersink: add AV_BUFFERSINK_FLAG_NO_REQUEST.
[14:30] <CIA-17> ffmpeg: 03Nicolas George 07master * r99f29108d3 10ffmpeg/ffmpeg.c: 
[14:30] <CIA-17> ffmpeg: ffmpeg: directly request frames from filters.
[14:30] <CIA-17> ffmpeg: It allows ffmpeg filtering logic to work with filters
[14:30] <CIA-17> ffmpeg: that do not implement the poll_frame method,
[14:30] <CIA-17> ffmpeg: such as split or tile.
[14:30] <CIA-17> ffmpeg: 03Nicolas George 07master * r2ce7972779 10ffmpeg/libavfilter/ (5 files): 
[14:30] <CIA-17> ffmpeg: avfiltergraph: add avfilter_graph_request_oldest().
[14:30] <CIA-17> ffmpeg: Keep a heap of all sink links ordered by timestamps.
[14:30] <CIA-17> ffmpeg: 03Nicolas George 07master * rf75ee4031e 10ffmpeg/libavfilter/vsrc_buffer.c: 
[14:30] <CIA-17> ffmpeg: vsrc_buffer: return EAGAIN if no frame is available.
[14:30] <CIA-17> ffmpeg: This is not an erroneous condition, do not print a warning.
[14:36] <ubitux> oh, awesome :)
[14:51] <CIA-17> ffmpeg: 03Nicolas George 07master * r9246896f77 10ffmpeg/ (3 files in 3 dirs): parseutils: add FATE test.
[14:51] <CIA-17> ffmpeg: 03Nicolas George 07master * r9b84f8a987 10ffmpeg/ (libavutil/parseutils.c tests/ref/fate/parseutils): parseutils: add av_parse_time() test.
[14:51] <CIA-17> ffmpeg: 03Nicolas George 07master * r43b7068070 10ffmpeg/libavutil/parseutils.c: 
[14:51] <CIA-17> ffmpeg: parseutils: make av_parse_time() check for failure.
[14:51] <CIA-17> ffmpeg: Until now, av_parse_time() would accept "1:00" as "1"
[14:51] <CIA-17> ffmpeg: and silently ignore ":00".
[14:51] <CIA-17> ffmpeg: This patch also includes a few cosmetic changes.
[18:16] <CIA-17> ffmpeg: 03Nicolas George 07master * r1e43786b4c 10ffmpeg/doc/print_options.c: print_options: warn that the files are generated.
[20:14] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r068d0b4e25 10ffmpeg/libavcodec/h264_sei.c: 
[20:14] <CIA-17> ffmpeg: h264: some fields in SEIs are longer than 25 bits thus use get_bits_long()
[20:14] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[20:14] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:15] <CIA-17> ffmpeg: 03Reimar Döffinger 07master * r53253a0b83 10ffmpeg/tests/ (Makefile fate/mapchan.mak): (log message trimmed)
[21:15] <CIA-17> ffmpeg: Simplify asynth sample generation.
[21:15] <CIA-17> ffmpeg: One rule can be used to generate all asynth files.
[21:15] <CIA-17> ffmpeg: Requires renaming the mapchan files though.
[21:15] <CIA-17> ffmpeg: Also switch to using the .wav variants for mapchan
[21:15] <CIA-17> ffmpeg: while changing the name anyway, this allows getting rid
[21:15] <CIA-17> ffmpeg: of the explicitly specified format.
[23:03] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r2954574126 10ffmpeg/libavcodec/indeo4.c: 
[23:03] <CIA-17> ffmpeg: indeo4: check quant_mat more fully.
[23:03] <CIA-17> ffmpeg: quant_mats valid range depends on the block size.
[23:03] <CIA-17> ffmpeg: This fixes a global array overread.
[23:03] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[23:03] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:26] <CIA-17> ffmpeg: 03Reimar Döffinger 07master * rc799e46231 10ffmpeg/tests/ (Makefile codec-regression.sh regression-funcs.sh): 
[23:26] <CIA-17> ffmpeg: Get rid of tests/data/asynth-%.sw rule.
[23:26] <CIA-17> ffmpeg: This also avoids an issue with parallel make in some
[23:26] <CIA-17> ffmpeg: cases never building asynth-16000-1.sw.
[23:26] <CIA-17> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[23:49] <CIA-17> ffmpeg: 03Nicolas George 07master * r2e8f19d26f 10ffmpeg/libavfilter/avfiltergraph.c: avfiltergraph: free the sink_links heap.
[00:00] --- Tue Apr 24 2012