[Ffmpeg-devel-irc] ffmpeg-devel.log.20120122

[burek]

[00:47] Action: Compn just realized its ffmpeg-devel at ffmpeg.org
[00:47] <Compn> not ffmpeg-devel at mplayerhq.hu :P
[00:47] Action: Compn goes back to doing whatever
[00:49] <ubitux> saste: i guess you are aware of the concat: protocol, doing the same thing than the cat x y z > out?
[00:50] <saste> ubitux: of course, after all it was me which committed/reviewed that protocol ;-)
[00:50] <ubitux> ok :D
[00:50] <ubitux> i had a doubt since you mentionned the cat command
[01:11] <saste> ubitux: nice work with the aresample
[01:11] <ubitux> it wasn't the harder i think ;)
[01:11] <ubitux> i was just trying to understand things about resampling
[01:11] <ubitux> and got into the filter
[01:12] <ubitux> so i just took a few minutes and give it a try with swr :p
[01:12] <saste> ubitux: i'll try to create a branch with the -af option in ffmpeg and some fate tests, so you can test that more extensively
[01:12] <Compn> someone should mail aurel and ask him bout mkv subs ;P
[01:12] <ubitux> oh yes fate tests will be very welcome
[01:12] <Compn> ooo audio filters ;)
[01:12] <saste> my wild guess is the recent rework in the audio API broke something
[01:12] <ubitux> arh subs, i need to work on them.
[01:13] <ubitux> saste: why? did you expect audio issues?
[01:13] <saste> yes there's something wrong in the lavfi/lavu interface
[01:13] Action: Compn waits for drv and his vivo demuxer
[01:14] Action: Compn campaigns for vivo support
[01:14] <saste> lavfi assumes a maximum number of planes (8), which is not anymore a constraint in the audio lavu utils 
[01:15] <ubitux> btw i'm still lost about the subtitles rework, i will have more time to work on them in a few weeks
[01:15] <saste> also now you can find planar/packed sample formats, while lavfi assumed a sample_fmt doesn't contain that info, so it uses a distinct flag (packed/planar)
[01:16] <ubitux> (got to get things done for school)
[01:16] <ubitux> yes the packed/planar is a bit inconsistent now
[01:16] <saste> indeed, i don't know maybe it's better that way, at least it is more consistent with pix_fmts
[01:18] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * rc5c7e0ebc1 10ffmpeg/ffmpeg.c: (log message trimmed)
[01:18] <CIA-31> ffmpeg: git show 2473a45c85dce6872617b33fce396dbbd6347e8e avconv.c | patch ffmpeg.c
[01:18] <CIA-31> ffmpeg: Original commit:
[01:18] <CIA-31> ffmpeg: commit 2473a45c85dce6872617b33fce396dbbd6347e8e
[01:18] <CIA-31> ffmpeg: Author: Janne Grunau <janne-libav at jannau.net>
[01:18] <CIA-31> ffmpeg: Date: Wed Jan 18 10:53:41 2012 +0100
[01:18] <CIA-31> ffmpeg:  threads: change the default for threads back to 1
[01:18] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r44d4fe9dad 10ffmpeg/ (libavcodec/options.c libavformat/options.c): 
[01:18] <CIA-31> ffmpeg: error_detection: put the old error detection constants back.
[01:18] <CIA-31> ffmpeg: The old constants are more useful to the end user and have
[01:18] <CIA-31> ffmpeg: orthogonal meanings to the new ones.
[01:18] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:18] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r2478612fc0 10ffmpeg/libavcodec/options.c: 
[01:18] <CIA-31> ffmpeg: lavc: Mark the err_detect constants as audio+video as they also apply to audio.
[01:18] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:18] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r657878f76e 10ffmpeg/libswscale/swscale_unscaled.c: 
[01:18] <CIA-31> ffmpeg: sws: fix copy case detection with gray8a.
[01:18] <CIA-31> ffmpeg: Fixes gray_alpha.png
[01:18] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:19] <ubitux> i'm a bit disappointed libav used a flag for the packed/planar
[01:19] <ubitux> i really like michael's idea with the flag bit in format
[01:19] <ubitux> but well, it's too late now :(
[01:19] <CIA-31> ffmpeg: img2: add j2c file extension
[01:19] <CIA-31> ffmpeg: Some applications use the j2c extension for jpeg2000 codestream files.
[01:19] <CIA-31> ffmpeg: Signed-off-by: Jean First <jeanfirst at gmail.com>
[01:19] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:20] <saste> ubitux: what was michaelni's idea?
[01:21] <ubitux> fmt & 0x100 -> packed
[01:21] <ubitux> planar*
[01:21] <saste> ok
[01:21] <ubitux> it was done like that in libswr at first
[01:22] <ubitux> (see 58205edcc67f606f13f2bc4571e5f5bd61eaef05)
[01:22] <saste> ah i see...
[01:23] <saste> time_to_sleep... good night all!
[01:23] <ubitux> also, in lavu/samplefmt.c, it even has a .planar flag while it could be done more nicely :(
[01:23] <ubitux> 'night saste :)
[02:07] <durandal_1707> heh i can not reproduce this: http://forum.doom9.org/showthread.php?p=1552295#post1552295
[02:08] <durandal_1707> it is bad because libav bugs are being considered as ffmpeg bugs too
[02:10] <Compn> its only going to get worse durandal_1707
[02:10] <Compn> bug dark_shikari about making x264 report if it was compiled with libav or ffmpeg
[02:11] <Compn> that might cut down on some crap
[02:12] <durandal_1707> this 48bit this is older, just looked post back in same thread
[02:24] <Compn> you mean old report ?
[02:25] <durandal_1707> no, similar issue with 48bit swscale this is reported in may
[02:25] <durandal_1707> in same thread
[04:49] <CIA-31> ffmpeg: 03Carl Eugen Hoyos 07master * rf746f37901 10ffmpeg/libavcodec/tiff.c: Support decoding gray8a tiff images.
[07:05] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r4f11bed7a1 10ffmpeg/libavformat/avidec.c: 
[07:05] <CIA-31> ffmpeg: avidec: fix "avidec.c:362: warning: st may be used uninitialized in this function"
[07:05] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[07:05] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r9e34adfc02 10ffmpeg/libavcodec/ratecontrol.c: 
[07:05] <CIA-31> ffmpeg: RC: Fix initial buffer occupancy.
[07:05] <CIA-31> ffmpeg: Fixes Ticket880
[07:05] <CIA-31> ffmpeg: Regression since bf174a70cde333eba7e27063f9ce44497108f466
[07:05] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[07:05] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r0421b6dc7a 10ffmpeg/libswscale/utils.c: 
[07:05] <CIA-31> ffmpeg: sws: In case of an invalid scaler algorithm, show the invalid value.
[07:05] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[07:05] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r887cd591a8 10ffmpeg/libavformat/mpegenc.c: 
[07:05] <CIA-31> ffmpeg: mpegpsmux: Fix integer overflow with huge VBV sizes.
[07:05] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[07:05] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r018a6645bc 10ffmpeg/libavcodec/aacenc.c: 
[07:05] <CIA-31> ffmpeg: aacenc: Fix bug in writing libavcodec_ident.
[07:05] <CIA-31> ffmpeg: the written length was off by 2 causing aac decoders to fail with the data.
[07:05] <CIA-31> ffmpeg: lucky the encoder was marked as experimental and not used much
[07:05] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[14:42] <CIA-31> ffmpeg: 03Reimar Döffinger 07master * rb09c93d7bc 10ffmpeg/libavcodec/fraps.c: 
[14:42] <CIA-31> ffmpeg: Move some duplicated code, simplify.
[14:42] <CIA-31> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[14:42] <CIA-31> ffmpeg: 03Reimar Döffinger 07master * rdaa47fdd2c 10ffmpeg/libavutil/base64.c: 
[14:42] <CIA-31> ffmpeg: Optimized base64 decode by writing 3 bytes at once.
[14:42] <CIA-31> ffmpeg: About 25% faster.
[14:42] <CIA-31> ffmpeg: decode: 248852 -> 200385 decicycles
[14:42] <CIA-31> ffmpeg: (syntax check unchanged)
[14:42] <CIA-31> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[14:42] <CIA-31> ffmpeg: 03Reimar Döffinger 07master * r420719e141 10ffmpeg/libavutil/base64.c: 
[14:42] <CIA-31> ffmpeg: Cosmetics: add do {} while 0 to macro.
[14:42] <CIA-31> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[14:42] <CIA-31> ffmpeg: 03Reimar Döffinger 07master * rb64b1608d8 10ffmpeg/libavcodec/fraps.c: 
[14:42] <CIA-31> ffmpeg: Remove pointless and distracting ()
[14:42] <CIA-31> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[14:42] <CIA-31> ffmpeg: 03Reimar Döffinger 07master * r57608d9029 10ffmpeg/libavcodec/fraps.c: 
[14:42] <CIA-31> ffmpeg: Simplify: just add header_size directly.
[14:42] <CIA-31> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[14:42] <CIA-31> ffmpeg: 03Reimar Döffinger 07master * rb4ec645f47 10ffmpeg/libavcodec/fraps.c: 
[14:42] <CIA-31> ffmpeg: Extract some duplicated code.
[14:42] <CIA-31> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[16:06] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * re9a6152e17 10ffmpeg/libavcodec/imgconvert.c: 
[16:06] <CIA-31> ffmpeg: lavc: fix FF_LOSS_COLORQUANT detection for gray8a->pal8
[16:06] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:06] <CIA-31> ffmpeg: 03Hendrik Leppkes 07master * r378b7beff5 10ffmpeg/libavcodec/mpeg12.c: 
[16:06] <CIA-31> ffmpeg: mpeg12: fixed parsing in some mpeg2 streams
[16:06] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:06] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r234405315f 10ffmpeg/libswscale/swscale_unscaled.c: 
[16:06] <CIA-31> ffmpeg: sws: Fix unscaled >8bit planar chroma handling.
[16:06] <CIA-31> ffmpeg: Fixes Ticket840
[16:06] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:16] <CIA-31> ffmpeg: 03Alexander Strasser 07master * r72b33eade1 10ffmpeg/configure: 
[17:16] <CIA-31> ffmpeg: build: Support selecting yasm-compatible assembler
[17:16] <CIA-31> ffmpeg:  Makes it possible to select the name/path of the tool for compiling
[17:16] <CIA-31> ffmpeg: the non-inline assembly code.
[17:16] <CIA-31> ffmpeg: Signed-off-by: Alexander Strasser <eclipse7 at gmx.net>
[17:16] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:39] <saste> is the libdvdnav  project still active?
[17:39] <saste> where can I find their devs?
[18:08] <Compn> saste : there is a mailing list for it i think
[18:08] <Compn> still alive ? maybe.
[18:08] <Compn> saste : maybe you should look into vlc's dvd reading code ?
[18:09] <Compn> mplayer hosts libdvdnav :P
[18:09] <Compn> Rathann : you still work on it ?
[18:10] <Compn> saste : http://lists.mplayerhq.hu/pipermail/dvdnav-discuss/2012-January/date.html
[18:10] <Compn> j-b is on there 
[18:10] <Compn> maybe vlc uses mplayer's forked libdvdnav
[18:13] <saste> Compn: thanks
[18:19] <Compn> our fork is still alive that is, the original looks dead
[18:19] <saste> uhm.. ok
[18:19] <saste> i'll go check the svn log
[18:21] <saste> erik, rathann, j-b, so it's still alive somehow
[18:22] <saste> looks like i need to do some changes for integrating the DVD protocol in ffmpeg (the old patch assumed some functions which are not implemented)
[18:22] <j-b> VLC uses libdvdav+dvdread from mplayerhq.hu
[18:22] <j-b> 4.2.0
[18:22] <saste> not that i trust so much the original contributor of the patch ;-)
[18:23] <saste> j-b: thx
[18:23] <j-b> because the fork is a bit less crap (but not that much)
[18:23] <j-b> dvdread and dvdnav code is so ugly that one's eyes bleed
[18:23] <saste> eheh people should be pretty used to that
[18:46] <NapHtaKeRoSene> hi
[18:46] <NapHtaKeRoSene> can somebody give a WORKING example of using gcc's inline asm in intel syntax? or it is not possible?
[18:47] <nevcairiel> gcc inline asm is at&t syntax
[18:47] <nevcairiel> you cant use intel syntax
[18:47] <iive> nevcairiel: i think that since gcc-3.x they introduced the intel syntax too
[18:48] <iive> it was one of the reasons for the 2.96.x (RH backported 3.0) to skip whole assembler blocks...
[18:49] <iive> but IMHO, i've never seen it used, either.
[18:50] <nevcairiel> quick googling shows that when using intel syntax you're somewhat limited
[19:02] <Compn> j-b : going to mentor SOC project to fix up dvdnav / dvdread ?
[19:02] <Compn> its used in at least two projects, and if ffmpeg joins that ...
[19:02] <Compn> i talk to users daily who say they have to use mencoder to get dvd input, then pipe it to ffmpeg :P
[19:03] <Compn> well i guess j-b is busy running like 10 diff projects already, nm :)
[19:04] <Compn> i liked that graph of linux video editors and how long they've lasted over the years :)
[19:13] Action: Compn wonders why he cant connect to ftp://upload.ffmpeg.org
[19:14] <Compn> j-b : streams.videolan.org down ?
[19:23] Action: Compn argues with patch authors on ffmpeg-devel
[19:23] <Compn> copyright concern trolling is having a chilling effect on development !!
[19:25] <nevcairiel> pgs is usually von blurays, he could just mention which disc. .)
[19:25] <nevcairiel> *from
[19:25] <nevcairiel> wrong language sneaking in
[19:29] <nevcairiel> I could probably find a disc in my library where this change helps, but thats effort
[20:25] <beastd> hi
[20:33] <Compn> nevcairiel : yeah he could mention the title
[22:47] <durandal_1707> looks like base64 is failing on big endian
[22:47] <Daemon404> ... shouldnt base64 go bye-by-byte
[22:47] <Daemon404> how does endian-ness even come into play?
[22:47] <nevcairiel> if you optimize it alot it might be an issue. :)
[22:48] <Daemon404> indeed\
[22:48] Action: durandal_1707 hopes this is gcc bug
[22:49] Action: Daemon404 does all his development with clang :3
[22:57] <durandal_1707> those code names are really ...
[23:08] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r2bb7396bfd 10ffmpeg/libavutil/base64.c: 
[23:08] <CIA-31> ffmpeg: base64: 10l endian fix.
[23:08] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:08] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * rc6baa4046a 10ffmpeg/libavfilter/avfilter.c: 
[23:08] <CIA-31> ffmpeg: avfilter: add some asserts() to check refcounts.
[23:08] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:43] <CIA-31> ffmpeg: 03Philip Langdale 07master * reaf4bf6df2 10ffmpeg/libavcodec/crystalhd.c: 
[23:43] <CIA-31> ffmpeg: CrystalHD: Initialise variables to silence valgrind.
[23:43] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:43] <CIA-31> ffmpeg: 03Alexander Strasser 07master * r9c909e8e79 10ffmpeg/configure: 
[23:43] <CIA-31> ffmpeg: build: Describe --nm and -ld options consistently
[23:43] <CIA-31> ffmpeg: Signed-off-by: Alexander Strasser <eclipse7 at gmx.net>
[23:43] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:43] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r89ddff92a3 10ffmpeg/ (6 files in 3 dirs): (log message trimmed)
[23:43] <CIA-31> ffmpeg: Revert "movenc: simplify handling of pcm vs. adpcm vs. other compressed codecs"
[23:43] <CIA-31> ffmpeg: This reverts commit 8e8c51318c1fe4ae61de578f0823b88aa3fe8222.
[23:43] <CIA-31> ffmpeg: Fixes Ticket873
[23:43] <CIA-31> ffmpeg: Conflicts:
[23:43] <CIA-31> ffmpeg:  tests/ref/acodec/pcm_s16be
[23:43] <CIA-31> ffmpeg:  tests/ref/acodec/pcm_s24be
[23:43] <CIA-31> ffmpeg: 03Philip Langdale 07master * r9bf9c314a0 10ffmpeg/libavcodec/crystalhd.c: (log message trimmed)
[23:43] <CIA-31> ffmpeg: CrystalHD: Back up extradata to allow decoder reinit to work.
[23:43] <CIA-31> ffmpeg: This was a regression that came in when I switched to using the
[00:00] --- Mon Jan 23 2012