[Ffmpeg-devel-irc] ffmpeg-devel.log.20111218

[burek]

[00:05] <redbal> rg, not getting a response with that command "awk --version",,,man page says SunOS 5.11...that' probably the culprit
[00:06] <redbal> I'll play around with a bit later...thanks michaelni
[01:08] <CIA-49> ffmpeg: 03Stefano Sabatini 07master * r86b7db616d 10ffmpeg/libavdevice/lavfi.c: lavd/lavfi: fix typo in error message
[01:09] <CIA-49> ffmpeg: 03Stefano Sabatini 07master * r21dfd1591e 10ffmpeg/libavdevice/lavfi.c: lavd/lavfi: fix incomplete comment
[01:14] <ubitux> oh, the typo i forgot to fix last time
[01:14] <ubitux> Compn: hey i was right about "output" ;)
[01:22] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r0db7b30700 10ffmpeg/libavcodec/lagarith.c: 
[01:22] <CIA-49> ffmpeg: lagarith: rgb24 support
[01:22] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:22] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r8beeec8821 10ffmpeg/libavcodec/sunrast.c: 
[01:22] <CIA-49> ffmpeg: sunrast: Fix warning: comparison of unsigned expression < 0 is always false [-Wtype-limits]
[01:22] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:23] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * rbd35dfeaa0 10ffmpeg/libavcodec/tta.c: 
[01:23] <CIA-49> ffmpeg: tta: Fix several warning: initialization from incompatible pointer type [enabled by default]
[01:23] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:23] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * rc9dc66375b 10ffmpeg/libavcodec/v210enc.c: 
[01:23] <CIA-49> ffmpeg: v210enc: Fix warning: val may be used uninitialized in this function [-Wuninitialized]
[01:23] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:23] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r53509d20ca 10ffmpeg/libavcodec/v210dec.c: 
[01:23] <CIA-49> ffmpeg: v210dec: Fix warning: val may be used uninitialized in this function [-Wuninitialized]
[01:23] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:49] <durandal_1707> there is some strange small mov on samples which keeps playing forever
[01:50] <durandal_1707> http://samples.libav.org/mov/protected_mov/ep2_mystery_m640_dummy.mov
[02:08] <Compn> durandal_1707 : a 'loop' 
[02:10] <Daemon404> pretty sure thats a valid feature of mov
[02:10] <Daemon404> iirc
[02:14] <CIA-49> ffmpeg: 03Michael Niedermayer 07release/0.9 * r7614facf8a 10ffmpeg/libavcodec/h264_mvpred.h: 
[02:14] <CIA-49> ffmpeg: h264: fix init of topleft ref/mv.
[02:14] <CIA-49> ffmpeg: Fixes Ticket778
[02:14] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:14] <CIA-49> ffmpeg: (cherry picked from commit 4d70023a2a83c5c7690708a5a5acfe8ac8ed8f19)
[02:33] <caytchen> so I'm wondering, are there actually decoders for every CODEC_ID?
[02:33] <durandal_1707> there is xmpg fourcc in avi
[02:37] <Daemon404> caytchen, i dont think so
[02:37] <Daemon404> i could be wrong.
[02:37] <caytchen> because there is a truly insane number of codecs there including some 1990 playstation games video codec.
[02:38] <Daemon404> caytchen, that particular one may actually exist
[02:40] <caytchen> there are a bunch of #if LIBAVCODEC_VERSION_MAJOR == 53 codecs in the midst, won't these break the ABI for versions > 53?
[02:46] <durandal_1707> isn't point of major bump to break ABI?
[03:12] <CIA-49> ffmpeg: 03Ronald S. Bultje 07master * r4391805916 10ffmpeg/libswscale/swscale.c: 
[03:12] <CIA-49> ffmpeg: swscale: fix overflows in RGB rounding constants.
[03:12] <CIA-49> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[03:12] <CIA-49> ffmpeg: 03Ronald S. Bultje 07master * r8cfbbd928c 10ffmpeg/ (configure libavcodec/get_bits.h libavcodec/wmavoice.c): (log message trimmed)
[03:12] <CIA-49> ffmpeg: get_bits: introduce safe bitreading to prevent overreads.
[03:12] <CIA-49> ffmpeg: When turned on, H264/CAVLC gets ~15% (CVPCMNL1_SVA_C.264) slower for
[03:12] <CIA-49> ffmpeg: ultra-high-bitrate files, or ~2.5% (CVFI1_SVA_C.264) for lower-bitrate
[03:12] <CIA-49> ffmpeg: files. Other codecs are affected to a lesser extent because they are
[03:12] <CIA-49> ffmpeg: less optimized; e.g., VC-1 slows down by less than 1% (all on x86).
[03:12] <CIA-49> ffmpeg: The patch generated 3 extra instructions (cmp, cmovae and mov) per
[03:12] <CIA-49> ffmpeg: 03Shitiz Garg 07master * re614fac2e6 10ffmpeg/libavcodec/adpcm.c: 
[03:12] <CIA-49> ffmpeg: adpcm: Check for channels to be a non-zero integer
[03:12] <CIA-49> ffmpeg: channels would be 0 sometimes and would cause floating point exception
[03:12] <CIA-49> ffmpeg: Fixes bugzilla #124
[03:12] <CIA-49> ffmpeg: Signed-off-by: Justin Ruggles <justin.ruggles at gmail.com>
[03:12] <CIA-49> ffmpeg: 03Mans Rullgard 07master * r583d3b7f97 10ffmpeg/tests/ (fate.mak ref/fate/ea-tqi-adpcm): 
[03:12] <CIA-49> ffmpeg: fate: fix eatqi test
[03:12] <CIA-49> ffmpeg: The sample has an incomplete last frame. Decoding it is pointless.
[03:12] <CIA-49> ffmpeg: The garbage produced was changed by the bitstream reader now
[03:12] <CIA-49> ffmpeg: protecting against over-reads.
[03:12] <CIA-49> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[03:13] <CIA-49> ffmpeg: 03Mans Rullgard 07master * rac08f798bb 10ffmpeg/libavcodec/mpegaudiodec.c: 
[03:13] <CIA-49> ffmpeg: mp3dec: hack: fix decoding with safe bitstream reader
[03:13] <CIA-49> ffmpeg: The buffer splicing relies on the bitstream reader over-reading
[03:13] <CIA-49> ffmpeg: the end of the buffer as declared in init_get_bits(), although
[03:13] <CIA-49> ffmpeg: more data is actually present. Manually moving the bitstream
[03:13] <CIA-49> ffmpeg: boundary after init_get_bits() allows this to work as expected.
[03:13] <CIA-49> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[03:13] <CIA-49> ffmpeg: 03Janne Grunau 07master * red46a3d842 10ffmpeg/libswscale/swscale_unscaled.c: swscale: add rgb565 endianess conversion
[03:13] <CIA-49> ffmpeg: 03Mans Rullgard 07master * re42bb0ee01 10ffmpeg/libavcodec/get_bits.h: 
[03:13] <CIA-49> ffmpeg: get_bits: whitespace (mostly) cosmetics
[03:13] <CIA-49> (57 lines omitted)
[03:15] <j-b> so, get_bits is in...
[03:15] <j-b> nice.
[03:39] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r38331d2036 10ffmpeg/libavcodec/ (h264.c h264_cabac.c h264_cavlc.c h264_parser.c): 
[03:39] <CIA-49> ffmpeg: h264: disable checking reader, overreads are not possible
[03:39] <CIA-49> ffmpeg: in ffmpegs h264 decoder.
[03:39] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:39] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r8e7776036b 10ffmpeg/libavcodec/ (h263dec.c ituh263dec.c): 
[03:39] <CIA-49> ffmpeg: h263: disable slow checked reader, overreads are not possible in
[03:39] <CIA-49> ffmpeg: ffmpegs h263 decoder
[03:39] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:39] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r06192e42fd 10ffmpeg/libavcodec/ (mpeg4video_parser.c mpeg4videodec.c): 
[03:39] <CIA-49> ffmpeg: mpeg4: disable checked bitstream reader, overreads are not possible.
[03:39] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:39] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r549749c68e 10ffmpeg/libavcodec/mpeg12.c: 
[03:39] <CIA-49> ffmpeg: mpeg12: disable checked reader.
[03:39] <CIA-49> ffmpeg: overreads in mpeg12 are not possible
[03:46] <michaelni> btw, j-b above commits also mean i consider any overread a serious bug in these decoders, so in the unlikely case i missed something and you spot them overreading, tell me and ill fix it
[03:47] <j-b> michaelni: ok
[03:47] <j-b> michaelni: I think trunk does not compile because of ELOOP
[03:48] <j-b> but, to get back to get_bits, I believe that security > speed those days
[03:52] <michaelni> j-b, which platform is it failing with ELOOP ?
[03:52] <michaelni> iam asking for the commit message
[03:52] <michaelni> better said, which platform is missing ELOOP
[03:53] <j-b> win32
[03:54] <michaelni> thx & thx for the report
[03:56] <michaelni> j-b, i of course agree that security is more important than speed, and the checked reader is also enabled by default
[03:56] <Daemon404> wee, ffmpeg builds for win32/win64 using clang cross-compiled
[03:56] Action: Daemon404 sets up some nightlies
[03:56] <michaelni> ... except for decoders where its known its unneeded
[03:57] <j-b> sure, sure
[03:58] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r4734fb0ece 10ffmpeg/libavformat/sbgdec.c: 
[03:58] <CIA-49> ffmpeg: sbgdec: Replace ELOOP by -1 to fix compilation on win32.
[03:58] <CIA-49> ffmpeg: This should be replaced by a more appropriate error code of course but
[03:58] <CIA-49> ffmpeg: we should not leave compilation broken until that is decided.
[03:58] <CIA-49> ffmpeg: Found-by: jb
[03:58] <j-b> understanding the mpeg* compiler is 1000 miles above my skillz
[03:58] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:05] Action: michaelni wonders what a mpeg* compile ris ;)
[04:10] Action: Daemon404 too
[04:11] <Daemon404> and sure is michaelni being up at 4am
[11:16] <Oana_> can a hang in ffmpeg be considered a bug?
[11:30] <Oana_> http://www.google-melange.com/gci/task/view/google/gci2011/7209215  can a mentor review this?
[14:04] <Oana__> can a mentor review http://google-melange.appspot.com/gci/task/view/google/gci2011/7209215 ?
[14:54] <michaelni> Oana_, sorry for the delay i was sleeping
[14:55] <michaelni> Oana_, its not hanging (infinite loop) nor crash (FPE/sig11) here
[14:57] <michaelni> OanaStratulat, actually, it is hanging , i was still sleepy :)
[14:57] <michaelni> sorry
[14:57] <michaelni> task is ok
[14:58] <OanaStratulat> ok 10x
[15:00] <OanaStratulat> is there another program like zzuf that modifies files for testing?
[15:02] <OanaStratulat> http://google-melange.appspot.com/gci/task/view/google/gci2011/7214217  please accept
[15:05] <michaelni> done
[15:05] <michaelni> and yes there are others
[15:05] <michaelni> i cant find a good list
[15:06] <michaelni> googling for "file fuzzer" and similar might work
[15:10] <Compnn> we should list them on the wiki ..
[15:10] <Compnn> reverse engineering page
[15:24] <michaelni> anyone wants to help review http://www.google-melange.com/gci/task/view/google/gci2011/7204204 ?
[15:24] <michaelni> its about logo / art
[15:29] <Compnn> ehe
[15:29] <Compnn> reminds me of photoshopped images from 1998
[15:30] <Compn> season of spring? hmm
[15:31] <Compn> i wonder if the artist could background the spring into the text parts, and drop the background
[15:31] <Compn> e.g. you could see the birds in the letters 
[15:31] <Compn> like google logos do i guess
[15:32] <michaelni> Compn, do you want to take over mentoring this one, iam like always quite busy ...
[15:32] <michaelni> also iam not sure what to awnser
[15:32] <buzz_> the logo is pretty enough, but perhaps too busy with the background. what compn said too
[15:32] <michaelni> his autumn logo, look quite nice
[15:33] <Compn> michaelni : sorry, i am busy planning for my trip or i would :(
[15:34] <buzz_> interesting idea to go for seasons, of course, will confuse visitors from the other side of the world ;-)
[15:35] <michaelni> i guess they are used to seeing the wrong season on US/EU pages being mentioned
[15:49] <CIA-49> ffmpeg: 03Nicolas George 07master * r6d7b50117a 10ffmpeg/libavformat/sbgdec.c: 
[15:49] <CIA-49> ffmpeg: sbgdec: change -1 into AVERROR(EINVAL).
[15:49] <CIA-49> ffmpeg: This is not as accurate as ELOOP, but there is an explicit error message
[15:49] <CIA-49> ffmpeg: just before anyway.
[16:00] <michaelni> Compn, buzz_, he uploaded a new different logo, comments ?
[16:01] <michaelni> comments from everyone else very welcome too ...
[16:02] <buzz_> i think it stands out better like that yeh. ie. enhances the logo
[16:04] <buzz_> what about a xmas logo with the libav logo buried in snow next to the ffmpeg logo ;-)
[16:04] <buzz_> heh
[16:06] <buzz_> i like themed logos anyway.. have done that for many years on my site - http://www.exotica.org.uk/wiki/ExoticA_Intro_Logos :)
[16:19] <Compn> buzz_ : i think we try not to flame libav guys
[16:19] <Compn> well i dunno the rules
[16:19] <Compn> and they call me a troll anyways ;)
[16:20] <buzz_> heh
[16:21] <ubitux> i also strongly disagree about flaming libav guys :)
[16:21] <ubitux> we have enough troubles :p
[16:21] <buzz_> i was only joking, but it's not a flame it is a friendly dig. like the old platform wars of the 1980s/1990s ;-)
[16:21] <Compn> michaelni : yes, the ffmpeg text logo is exactly what i was looking at
[16:22] <Compn> and we dont need those birds... :)
[16:22] <Compn> er looking for
[16:22] <Compn> looks good.
[16:22] Action: Compn goes afk
[16:22] <Compn> spring12 that is
[16:23] <ubitux> < buzz_> i was only joking, but it's not a flame it is a friendly dig. // most of them take these kind of jokes too seriously ;)
[16:23] <buzz_> yeh. well i gathered the level of humour when ffmpeg got the notice regarding logo use
[16:24] <buzz_> how nice
[16:33] <michaelni> anyone knows of a list of free photos that our artist can use to make the logo without us getting into copyright troubble later ?
[16:36] <ubitux> iirc flickr has some license filtering
[16:37] <ubitux> http://animalphotos.info/a/
[16:37] <ubitux> "Images on this site are under creative commons and are free to use on web sites and other projects."
[16:37] <ubitux> just animals though
[16:37] <ubitux> i think i have a few more links somewhere, just a sec
[16:38] <ubitux> http://compfight.com/  the license is displayed when you select one
[16:38] <ubitux> http://www.flickr.com/creativecommons/  flickr cc
[16:39] <ubitux> http://openclipart.org/  public domain
[16:40] <michaelni> ok, let me copy and paste that ...
[16:43] <ubitux> http://photoeverywhere.co.uk/
[16:43] <ubitux> not sure how much "free" it is
[16:43] <ubitux> http://creativity103.com/about.htm  maybe this one too
[16:48] <ubitux> i think that's all i have from a rapid look up in my links
[17:07] <michaelni> btw, ubitux thx for the links
[17:08] <ubitux> np :)
[17:20] <ubitux> i mentionned a while ago the need of various meta information being raised to the filters
[17:20] <ubitux> so we could print them on the video
[17:20] <ubitux> http://hacks.mozilla.org/2011/12/new-features-for-html5-video-playback-in-firefox/   it seems mozilla did it for their player
[17:20] <ubitux> (2nd demonstrated feature)
[17:27] <ohsix> ffdshow has an overlay element that does a bunch of stuff too
[17:29] <ohsix> i use it a lot when i'm encoding stuff
[17:30] <ohsix> (you can set it to be placed as a "raw" filter, and media player classic can add filters to anything played with it)
[18:41] <ubitux> another kind of fuzzer: http://blitiri.com.ar/p/libfiu/
[18:41] <ubitux> it might help for GCI
[18:43] <ohsix> fiu is neat
[18:47] <ohsix> beats writing your own anyways
[20:36] <Cosmin> http://www.google-melange.com/gci/task/view/google/gci2011/7209212   can a mentor review ? the files that must be reviewd are lastfile and lastreport
[21:25] <michaelni> Cosmin, good work, infinite loop confirmed
[21:29] <Cosmin> thx
[21:36] <OanaStratulat> http://www.google-melange.com/gci/task/view/google/gci2011/7214217  marked as completed 
[21:48] <OanaStratulat> michaelni: is it good ?
[21:49] <michaelni> OanaStratulat, it seems to be the same bug Cosmin just found
[21:49] <michaelni> thats bad luck, iam myself not happy :(
[21:50] <OanaStratulat> oh, ok i`ll search for another one 
[21:54] <OanaStratulat> michaelni: http://pastebin.com/zZWVcPDg can this output of a file be considered a bug? in zzuf it gives an memory exceeded and i have 4gb ram an 1TB hdd
[21:54] <OanaStratulat> i`m running the latest ffmpeg version
[21:57] <michaelni> OanaStratulat, if you run it under valgrind its memory useage will be displayed
[21:58] <michaelni> from the pastebin i cant see how much it used
[21:58] <michaelni> the "total heap usage:" of valgrind
[21:59] <OanaStratulat> michaelni: http://pastebin.com/D3cLPYZV
[22:00] <michaelni> from the valgrind it looks like 1gb which is enough to qualify as bug when i can reproduce it (your version of ffmpeg is a few hours behind)
[22:01] <OanaStratulat> have you modified anything in the src code that involves memory usage?
[22:01] <OanaStratulat> i will post the files and see if you can reproduce it
[22:03] <OanaStratulat> http://www.google-melange.com/gci/task/view/google/gci2011/7214217   check it, use corruptfile2 and valgrind2
[22:06] <michaelni> OanaStratulat, reproduced, bug ok
[22:06] <OanaStratulat> michaelni: ok thx
[22:11] <OanaStratulat> if a valgrind report says at the leak summary that some bytes are definitely lost is that considered a bug?
[22:15] <michaelni> well, a memory leak is a bug but id say it should be a sigificant amount, or more precissely a growing amount
[22:15] <michaelni> i mean 300bytes lost independant of time file size and file count is not something any user would ever notice
[22:16] <OanaStratulat> my file when it is run in ffmpeg it gives a segfault but in valgrind there is no error
[22:16] <OanaStratulat> http://google-melange.appspot.com/gci/task/view/google/gci2011/7211207 please accept this task
[22:16] <michaelni> done
[22:17] <OanaStratulat> this is the output of ffmpeg http://pastebin.com/12E72Wbw   ; and this is the one from valgrind http://pastebin.com/rbsXdRCb
[22:19] <michaelni> if i can reproduce the crash then its ok ...
[22:19] <OanaStratulat> ok
[22:20] <OanaStratulat> http://google-melange.appspot.com/gci/task/view/google/gci2011/7211207   marked as completed
[22:25] <michaelni> it doesnt crash for me :(
[22:25] <OanaStratulat> bad for me
[22:25] <michaelni> not without valgrind not with nor with gdb
[22:26] <OanaStratulat> to get the last version of ffmpeg i have to clone the git again and ./conf and make?
[22:26] <michaelni> git pull --rebase
[22:26] <michaelni> will update it
[22:26] <michaelni> then ./conf and make
[22:27] <OanaStratulat> ok
[22:27] <OanaStratulat> maybe this is the prob that i`m getting bugs
[22:28] <OanaStratulat> how can i download all the samples from mplayer website faster? is ther an automatic script to dwld all the files on the website?
[22:28] <ohsix> they say not to do that bro
[22:28] <ohsix> also i'm pretty sure the same thing that says not to do that, says to contact them for alternative ways to get the samples
[22:29] <ohsix> though since the fork i dunno what's going on
[22:29] <ohsix> i guess nothing changed that mattered though
[22:29] <OanaStratulat> i want to test the files with ffmpeg because the fate system from libav is almost complete
[22:29] <nevcairiel> you know, ffmpeg also has fate
[22:30] <OanaStratulat> i did not know that
[22:30] <OanaStratulat> how can i get the samples?
[22:30] <Compn> its basically the same project
[22:30] <Compn> ffmpeg and libav i mean
[22:30] <Compn> should be a fate document in the documentation
[22:30] <OanaStratulat> yes, many bugs from libav are common in ffmpeg
[22:38] <OanaStratulat> michaelni: yes that was the problem, it does not sefgault anymore now
[22:43] <OanaStratulat> http://google-melange.appspot.com/gci/task/view/google/gci2011/7211207 please review now, use corruptfile 2 and valgrind 2 
[22:49] <OanaStratulat> michaelni: any luck with it?
[22:50] <michaelni> yes, looks ok
[22:51] <OanaStratulat> http://google-melange.appspot.com/gci/task/view/google/gci2011/7210227 please accept
[23:10] <OanaStratulat> michaelni: floating point exception is a bug?
[23:10] <michaelni> yes
[23:11] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7210227 can you review this
[23:11] <OanaStratulat> use the corruptfile2 and valgrind2
[23:14] <michaelni> OanaStratulat, there are no files attached also please create bugreports on ffmpeg.org/trac 
[23:15] <michaelni> without these bugreports we might forget about one of these bugs ...
[23:15] <OanaStratulat> michaelni: ok
[23:21] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7210227 please review
[23:33] <michaelni> OanaStratulat, done, good work
[23:34] Action: michaelni wonders if CIA-49 is sleeping
[23:40] Action: kierank kicks CIA-49
[23:40] <CIA-49> ow
[23:48] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7210232 please accept
[23:51] <burek> is there a way to compile one big fat static ffmpeg binary with all the codesc/formats/etc?
[23:51] Action: michaelni wants to add a "write a gci task accepter bot task"
[23:51] <michaelni> OanaStratulat, done
[23:53] <michaelni> burek, iam not sure i understand, static libavcodec/... is the default
[23:57] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7210232  please review this task
[00:00] --- Mon Dec 19 2011