[Ffmpeg-devel-irc] ffmpeg-devel.log.20120805

burek burek021 at gmail.com
Mon Aug 6 02:05:03 CEST 2012


[00:40] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r66747a6160 10ffmpeg/doc/examples/muxing.c: 
[00:40] <CIA-41> ffmpeg: examples/muxing: simplify alloc_picture()
[00:40] <CIA-41> ffmpeg: Use avpicture_alloc() high level function.
[00:40] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * re408e329d2 10ffmpeg/libavformat/img2dec.c: 
[00:40] <CIA-41> ffmpeg: lavf/img2dec: rename find_image_range() "max_start" parameter to "start_index"
[00:40] <CIA-41> ffmpeg: The new name seems more consistent with the assumed logic.
[00:40] <CIA-41> ffmpeg: "start_index" represents the minimum accepted value as first index, and
[00:40] <CIA-41> ffmpeg: not the maximum value as implicitely assumed by the previous name.
[00:40] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * ra30e776ff1 10ffmpeg/libavformat/img2dec.c: 
[00:40] <CIA-41> ffmpeg: lavf/img2dec: fix weird start_number count logic
[00:40] <CIA-41> ffmpeg: This seems consistent with the assumed meaning of the option.
[00:40] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r59e4e40673 10ffmpeg/libavformat/img2dec.c: lavf/img2dec: doxygen find_image_range() function
[00:40] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r48c1d1d3fa 10ffmpeg/libavformat/img2dec.c: lavf/img2dec: introduce and use FIRST_INDEX_SEARCH_RANGE macro in find_image_range
[00:40] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r2dfee8abf1 10ffmpeg/libavformat/img2dec.c: lavf/img2dec: sort options by name
[00:40] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * rb5c7318bd7 10ffmpeg/libavformat/img2dec.c: lavf/img2dec: improve error logging in case of find_image_range() failure
[00:40] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r07ef7b1a30 10ffmpeg/libavformat/img2dec.c: 
[00:40] <CIA-41> ffmpeg: lavf/img2dec: fix out-of-range check in find_image_range()
[00:40] <CIA-41> ffmpeg: Take the new parameter start_index into account.
[00:56] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * rb0d23ae208 10ffmpeg/ (doc/demuxers.texi libavformat/img2dec.c): lavf/img2doc: document options
[00:56] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r74c6db0505 10ffmpeg/doc/demuxers.texi: 
[00:56] <CIA-41> ffmpeg: doc/demuxers.texi: avoid comma splice in image2 documentation
[00:56] <CIA-41> ffmpeg: Add "and" proposition for connecting two sentences.
[05:12] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r9e1c55cfde 10ffmpeg/libavformat/oggdec.c: 
[05:12] <CIA-41> ffmpeg: oggdec: check stream index before using it in ogg_get_length()
[05:12] <CIA-41> ffmpeg: Fixes crash based on a uninitialized array index read.
[05:12] <CIA-41> ffmpeg: If the read does not crash then out of array writes based
[05:12] <CIA-41> ffmpeg: on the same index might have been triggered afterwards.
[05:12] <CIA-41> ffmpeg: Found-by: inferno at chromium.org
[05:12] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[10:24] <CIA-41> ffmpeg: 03Hendrik Leppkes 07master * r2db097ca76 10ffmpeg/configure: 
[10:24] <CIA-41> ffmpeg: configure: fix lib.exe check
[10:24] <CIA-41> ffmpeg: Running lib.exe without parameters will always result in an error,
[10:24] <CIA-41> ffmpeg: and failing the check.
[10:24] <CIA-41> ffmpeg: Using the -list paramter results in a success if lib.exe is present.
[10:58] <burek> is there a way to see what vcodec will ffmpeg choose by default, for a given output format?
[11:10] <saste> burek: no, but you should be able to write a small program to do so
[11:11] <saste> or we could extend the output of -formats to show that
[11:11] <burek> I was afraid you're gonna say that :)
[11:11] <saste> or add a show_formats option in ffprobe...
[11:11] <burek> I was thinking more like, is there a table in the source code that shows it :)
[11:12] <saste> no but is specified in each AVCodec structure, so once you register all codecs you can list them programmatically
[11:13] <burek> I see, well extending -formats is a cool idea
[15:49] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r9bece76009 10ffmpeg/ (3 files in 2 dirs): lavf/img2dec: add start_number_range option
[15:49] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * rb907655303 10ffmpeg/doc/examples/muxing.c: examples/muxing: drop duplicated code in add_video_stream()
[15:49] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r8d6f46ab2c 10ffmpeg/doc/filters.texi: 
[15:49] <CIA-41> ffmpeg: doc/filters: add mandelbrot source documentation
[15:49] <CIA-41> ffmpeg: Address trac ticket #1492.
[15:51] <burek> there are people at ffsplit.com that use ff* branding and actually use ffmpeg source code but they don't provide source code in their download
[15:51] <burek> just .exe file
[15:51] <burek> is that a kind of violation of ffmpeg's license
[16:13] <ubitux> they are on irc (quakenet)
[16:13] <ubitux> in case we would like to tell them to fix that asap.
[16:20] <ubitux> i'm not sure about the branding btw
[16:24] <ubitux> burek: there is a link to FFmpeg on the bottom of the download page
[16:25] <burek> well yes, I saw that, but if they use ffmpeg's source code and they are extending it in some way, isn't it mandatory that they respect the same license for their code?
[16:25] <burek> isn't that the purpose of GPL
[16:26] <burek> so that noone can close the source code and say "no, I added some important updates which I can't share"
[16:26] <nevcairiel> under the LGPL they are free to use it from a closed-source project, they just have to share any changes they do
[16:27] <burek> ffmpeg is LGPL?
[16:27] <nevcairiel> if all the extra stuff is in their code, and not a change in ffmpeg itself, they are free to not share
[16:27] <ubitux> burek: yes it's LGPL, some part are GPL though
[16:27] <ubitux> depends how they distribute the built one
[16:28] <ubitux> i mean, if they have --enable-gpl in they distributed libs it might be a problem
[16:28] <burek> ok, I'm not a legal expert, I just pointed the attention to the right direction, so you (or anyone who cares about it) can investigate if there is a need :)
[16:28] <ubitux> i'm not either
[16:49] <Compn> basically, you ask them for source
[16:49] <Compn> if they dont provide it, they arent following license
[17:16] <michaelni> hey saste, did you break fate ?
[17:16] <michaelni> fate-sub-movtextenc is failing here since 
[17:17] <michaelni> 9bece760097bf1de120e47aa504c9a9ca54d4490
[17:24] <saste> michaelni: how's that related to image2??
[17:27] <michaelni> its not ok to break fate even if the relation isnt obvious (its about bumping ver & bitexact here)
[17:28] Action: michaelni is pissed
[17:28] <CIA-41> ffmpeg: 03Philip Langdale 07master * r3eb56e8434 10ffmpeg/libavformat/movenc.c: 
[17:28] <CIA-41> ffmpeg: movenc: Fix unfreed memory found by valgrind.
[17:28] <CIA-41> ffmpeg: Signed-off-by: Philip Langdale <philipl at overt.org>
[17:28] <saste> is that supposed to happen? I mean, do we have tests which changes according to version bumps?
[17:30] <saste> I bumped just before to push, as it is my custom
[17:30] <michaelni> you should bump before testing :)
[17:31] <michaelni> and i would have fixed it already but -flags +bitexact isnt enough
[17:31] <ubitux> huh wtf
[17:31] <michaelni> somethings buggy so the flag isnt reaching the muxer with subtitles only 
[17:31] <ubitux> the micro bump changes the tests? oO
[17:32] Action: ubitux is also generally testing before bumping and might change his habit
[17:34] Action: michaelni thinks it would be best if people would send him pull requests instead of direct pushes, as iam able to run fate before push and itz would also catch files that where forgotten to be added but are local (like yesterdays header)
[17:35] <ubitux> i wonder if philip's last change is ok&
[17:36] <durandal_1707> michaelni: what kind of fate? you mean all machines on fate server?
[17:36] <michaelni> just mine
[17:36] <ubitux> mmh actually it's fine.
[17:37] <durandal_1707> imho, people who push stuff are expected to test them
[17:37] <michaelni> yes
[17:38] <ubitux> anyway, does anyone has an explanation for the failure?
[17:39] <saste> ubitux: the output file contains libavformat version in metadata
[17:39] <ubitux> oh
[17:39] <saste> as I bumped micro the hash changed
[17:39] <ubitux> interesting
[17:39] <durandal_1707> there was strange paf failure on one of machines when i fixed code and that machines resolved it after some time, so there are strange cases ....
[17:39] <saste> so anyway, I'm supposed to fix the reference?
[17:40] <durandal_1707> proper fix is that libavf version is not put into files
[17:41] <michaelni> that needs bitexact flag to be set 
[17:41] <michaelni> but for some reason it doesnt work with subtitles
[17:41] <durandal_1707> there should be flag in muxed/encoder to not put numbers if bitexact is set or something ...
[17:41] <michaelni> there is but it works with A/V but not S
[17:41] <durandal_1707> ffmpeg bug?
[17:46] <ubitux> saste: btw doc//filters.texi:3689: Bad argument `table.' to `@end', using `table'
[17:47] <ubitux> hey philipl :)
[17:47] <philipl> hi
[17:48] <saste> ubitux: i suppose it is because of the ending ".", weirdly it is not complaining here
[17:49] <ubitux> yes most likely
[17:52] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r8331d56e0d 10ffmpeg/doc/filters.texi: 
[17:52] <CIA-41> ffmpeg: doc/filters: fix typo in "@end table."
[17:52] <CIA-41> ffmpeg: Remove erroneously added trailing dot.
[17:52] <ubitux> saste: thx :)
[17:59] <michaelni> saste, i think ive fixed the bitexact bug
[18:00] <michaelni> ill push in a moment if it passes tests
[18:02] <philipl> :-( movtextenc test is now not passing. but the md5sum from my manual run of the command line still matches the md5sum I pushed originally.
[18:02] <ubitux> that's what michaelni was talking about
[18:02] <ubitux> it's about the encoder version being in the output
[18:02] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rb11b0e166e 10ffmpeg/libavcodec/options_table.h: 
[18:02] <CIA-41> ffmpeg: libavcodec/options_table: mark flags as also for subtitles
[18:02] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:03] <CIA-41> ffmpeg: 03Moritz Bunkus 07master * rbc3b422029 10ffmpeg/libavformat/ (matroska.c matroskadec.c): (log message trimmed)
[18:03] <CIA-41> ffmpeg: matroskadec: Implement support for ALAC
[18:03] <CIA-41> ffmpeg: This patch implements support reading ALAC from Matroska files. The
[18:03] <CIA-41> ffmpeg: only non-trivial thing about it is that only the ALAC magic cookie is
[18:03] <CIA-41> ffmpeg: stored in Matroska's CodecPrivate element but not the "atom size",
[18:03] <CIA-41> ffmpeg: "tag" and "tag version" fields that FFMPEG's ALAC decoder
[18:03] <CIA-41> ffmpeg: expects. However, those are trivial to re-create.
[18:03] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rd106ffd942 10ffmpeg/tests/ (fate/subtitles.mak ref/fate/sub-movtextenc): 
[18:03] <CIA-41> ffmpeg: fate: add bitexact flag to fate-sub-movtextenc
[18:03] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:03] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rc30f53e4d8 10ffmpeg/libavformat/matroskadec.c: 
[18:03] <CIA-41> ffmpeg: matroskadec: prevent potential integer overflow
[18:03] <CIA-41> ffmpeg: Iam not sure this can happen, but its safer we check.
[18:03] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:03] <philipl> Did he just save me?
[18:03] <ubitux> oh that was the 'S' flag missing
[18:03] <philipl> wrt bitexact?
[18:03] <philipl> Ok.
[18:03] <ubitux> i checked the constant and it had the 'S'
[18:04] <ubitux> but not the option.. :p
[18:04] <michaelni> yes :)
[18:04] <ubitux> philipl: bitexact prevents the encoder version from being there
[18:04] <ubitux> michaelni: well spotted :p
[18:04] <philipl> thanks.
[18:04] <philipl> I've posted a review to fix the ass valgrind warnings.
[18:04] <philipl> It's clearly dangerous to exercise any of that code in fate :-)
[18:06] <ubitux> philipl: i think it would be better to use av_freep()
[18:07] <philipl> ok.
[18:14] <CIA-41> ffmpeg: 03Philip Langdale 07master * r50b4dbf65a 10ffmpeg/ (ffmpeg.c libavcodec/ass_split.c): (log message trimmed)
[18:14] <CIA-41> ffmpeg: ass subtitles: Fix valgrind warnings.
[18:14] <CIA-41> ffmpeg: We're now running some of this code through valgrind for the first
[18:14] <CIA-41> ffmpeg: time, and a few warnings showed up stemming from two problems.
[18:14] <CIA-41> ffmpeg: 1) The ASS code assumes the subtitle header is null terminated, but
[18:14] <CIA-41> ffmpeg: it wasn't, and passing the size down doesn't look like fun, so I
[18:14] <CIA-41> ffmpeg: added a terminator
[18:17] <philipl> Off I go. Thanks gents.
[18:17] <ubitux> cya ;)
[18:34] <durandal_1707> why wav muxer doesn't write WaveFormatExtensible stuff?
[18:34] <CIA-41> ffmpeg: 03Moritz Bunkus 07master * rfe40dc1cec 10ffmpeg/libavformat/matroskaenc.c: matroskaenc: Implement support for ALAC
[18:53] <durandal_1707> i gonna push libtwolame
[18:57] <durandal_1707> and id3 tags should be written for mp2 files too
[19:05] <durandal_1707> what does I flag means unlike E it is not documented at all
[19:08] <CIA-41> ffmpeg: 03Paul B Mahol 07master * rf5f98727b3 10ffmpeg/ (6 files in 3 dirs): 
[19:08] <CIA-41> ffmpeg: libtwolame MP2 encoding support
[19:08] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[19:48] <CIA-41> ffmpeg: 03Reimar Döffinger 07master * ra4163b2d65 10ffmpeg/libavformat/oggdec.c: 
[19:48] <CIA-41> ffmpeg: oggdec: simplify start time calculation code.
[19:48] <CIA-41> ffmpeg: Also slightly more correct behaviour in case streams_left for
[19:48] <CIA-41> ffmpeg: some reason is 0 from the start.
[19:48] <CIA-41> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[19:48] <CIA-41> ffmpeg: 03Reimar Döffinger 07master * r69aeba1396 10ffmpeg/libavformat/oggdec.c: (log message trimmed)
[19:48] <CIA-41> ffmpeg: oggdec: Initialize stream index to -1 in ogg_packet.
[19:48] <CIA-41> ffmpeg: The previous method of having to initialize it outside lead
[19:48] <CIA-41> ffmpeg: to incorrect code: even if it was initialized, it usually was
[19:48] <CIA-41> ffmpeg: only initialized once, thus a packet that could not be matched
[19:48] <CIA-41> ffmpeg: to any stream would just be processed with the return values
[19:48] <CIA-41> ffmpeg: from the previous call.
[19:59] <durandal_1707> hmm why exr is in riff.c are there actual files?
[21:08] Action: ubitux finally got some csa reports
[21:13] <ubitux> but i have various crashes though.
[23:06] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * r29facc1e91 10ffmpeg/libavcodec/zerocodec.c: 
[23:06] <CIA-41> ffmpeg: zerocodec: Cosmetics
[23:06] <CIA-41> ffmpeg: Be consistent with error messages and code formatting.
[23:06] <CIA-41> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[23:06] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * r616fd4fe5e 10ffmpeg/libavcodec/zerocodec.c: 
[23:06] <CIA-41> ffmpeg: zerocodec: Fix memleak in decode_frame
[23:06] <CIA-41> ffmpeg: If there was a failure inflating, or reinitializing
[23:06] <CIA-41> ffmpeg: the zstream, the current frame's buffer would be lost.
[23:06] <CIA-41> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[23:06] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * r8b8750e061 10ffmpeg/ (libavcodec/cllc.c libavfilter/vf_libopencv.c): 
[23:06] <CIA-41> ffmpeg: cosmetics: Fix a few switched periods and linebreaks
[23:06] <CIA-41> ffmpeg: Based on a patch by Piotr Bandurski.
[23:06] <CIA-41> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[23:06] <CIA-41> ffmpeg: 03Reinhard Tartler 07master * r800750417f 10ffmpeg/libavfilter/audio.c: 
[23:06] <CIA-41> ffmpeg: lavfi: properly signal out-of-memory error in ff_filter_samples
[23:06] <CIA-41> ffmpeg: Found with a clang-scan report on http://fate.libav.org/csa/
[23:06] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * r965efc1673 10ffmpeg/libavcodec/dpx.c: 
[23:06] <CIA-41> ffmpeg: dpx: Make start offset unsigned
[23:06] <CIA-41> ffmpeg: Some corrupted files would end up with a negative offset,
[23:06] <CIA-41> ffmpeg: and segfault.
[23:07] <CIA-41> ffmpeg: rational: add av_inv_q() returning the inverse of an AVRational
[23:07] <CIA-41> ffmpeg: This allows simplifying a few expressions.
[23:07] <CIA-41> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[23:07] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rb4780d03d0 10ffmpeg/: (log message trimmed)
[23:07] <CIA-41> ffmpeg: Merge remote-tracking branch 'qatar/master'
[23:07] <CIA-41> ffmpeg: * qatar/master:
[23:07] <CIA-41> ffmpeg:  x86: h264_idct: Rename x264_add8x4_idct_sse2 --> h264_add8x4_idct_sse2
[23:07] <CIA-41> ffmpeg:  rational: add av_inv_q() returning the inverse of an AVRational
[23:07] <CIA-41> ffmpeg:  dpx: Make start offset unsigned
[23:07] <CIA-41> ffmpeg:  lavfi: properly signal out-of-memory error in ff_filter_samples
[23:07] <CIA-41> ffmpeg: 03Diego Biurrun 07master * r2096857551 10ffmpeg/libavcodec/x86/h264_idct.asm: x86: h264_idct: Rename x264_add8x4_idct_sse2 --> h264_add8x4_idct_sse2
[23:14] <j-b> lo
[23:14] <j-b> what is 0RGB fmt?
[23:15] <durandal_1707> similar to ARGB but A is always FF
[23:17] <j-b> ok
[23:19] <durandal_1707> looks like it was created just because of bmp
[23:20] <j-b> durandal_1707: fun
[23:30] <durandal_1707> j-b: why is that funny?
[23:33] <j-b> durandal_1707: because ;)
[23:43] <kierank> lol
[23:44] <j-b> because of the G2M patch
[23:45] <durandal_1707> there is G2M patch?
[23:49] <j-b> yes
[23:49] <durandal_1707> very bad joke, btw
[23:50] Action: kierank doesn't quite get the joke
[23:52] <durandal_1707> it is decoder that does almost nothing
[23:53] <durandal_1707> and kostya is almost begging that it get info ffmpeg
[23:53] <ubitux> haha
[23:54] <ubitux> refering to his latest blog post? :)
[23:54] <iive> hum? why not get it in libav?
[23:55] <ubitux> because it serves no purpose© i guess
[23:56] <ubitux> http://codecs.multimedia.cx/?p=446
[23:58] <CIA-41> ffmpeg: 03Piotr Bandurski 07master * r5d2f89a23c 10ffmpeg/libavformat/ (mov.c movenc.c): movenc: fix remuxing of svq3
[23:58] <iive> that blog entry doesn't really explain anything. is go2meeting some skype like software that never made it?
[00:00] --- Mon Aug  6 2012


More information about the Ffmpeg-devel-irc mailing list