[Ffmpeg-devel-irc] ffmpeg-devel.log.20111219

[burek]

[00:03] <burek> michaelni, I was thinking of building one big linux binary, just like windows has .exe
[00:04] <burek> for those who don't know how to compile or how to use package management
[00:04] <burek> so they can just download and run
[00:08] <OanaStratulat> michaelni: any luck with the bug?
[00:09] <michaelni> OanaStratulat, yes confirmed 
[00:10] <michaelni> burek, that sounds cool
[00:10] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7227202 please accept
[00:11] Action: durandal_1707 wonders why ffplay can not play smjpeg without additional arguments
[00:13] <michaelni> what arguments ?
[00:15] <burek> michaelni, are there any docs how could i do that?
[00:16] <durandal_1707> michaelni: no arguments at all
[00:17] <durandal_1707> michaelni: it appears to check extension only
[00:17] <durandal_1707> so if file is without extension it will mis detect
[00:17] <michaelni> burek, "ldd ffmpeg" will tell you what shard libs it links to, thats not an awnser i know ...
[00:18] <michaelni> ffmpeg will do static linking to its own libs by default
[00:18] <durandal_1707> burek: --extra-libs=-static --extra-cflags=-static
[00:18] <michaelni> durandal_1707, thx :)
[00:19] <burek> :)
[00:19] <durandal_1707> --enable-static --disable-shared
[00:19] <burek> and is there, by any chance, an option to --enable-all-codecs :)
[00:19] <durandal_1707> but i do not have that, so probably not needed
[00:19] <burek> btw thx durandal_1707  :)
[00:19] <durandal_1707> burek: that is by default
[00:20] <burek> i think the default is to check if the lib is there, then enable the support for it, otherwise silently skip it (disable support for it), right?
[00:20] <michaelni> about the file extension, its probably img2 not having a probe function 
[00:20] <durandal_1707> michaelni: smjpeg is mjpeg, not image
[00:21] <michaelni> hmm
[00:21] <durandal_1707> ffplay -f mjpeg br_06
[00:22] <michaelni> -f img2pipe doesnt work ?
[00:22] <michaelni> not that it matters
[00:22] <durandal_1707> burek: you can not distribute ffmpeg which is build with GPL: lame,x264
[00:23] <michaelni> either of them need a probe function that works
[00:24] <burek> can't even leave a link for people to freely download the big binary for their own use?
[00:24] <durandal_1707> michaelni: -f image2pipe reports could not open codecs
[00:25] <michaelni> maybe needs -vcodec mjpeg too
[00:25] <michaelni> durandal_1707, x264 is GPL
[00:25] <burek> I mean, does it mean they can compile it for their own use, but I can't compile it (for them) and just give them the download link, so they don't have to do all that themselves?
[00:25] <michaelni> and my memory says lame is GPL compatible
[00:26] <durandal_1707> burek: you are not allowed to distribute them
[00:26] <burek> that's just wrong..
[00:26] <durandal_1707> binaries
[00:27] <durandal_1707> you get that when you do ./configure --enable-libx264 ....
[00:27] <burek> can't wait for rothschilds to get arrested finally and get rid of all that patent crap
[00:27] <michaelni> lame is LGPL, x264 is GPL you can distribute ffmpeg linked to them 
[00:28] <michaelni> there are problems with other things like libfaac
[00:28] <burek> well ffmpeg does have an internal aac, right?
[00:28] <durandal_1707> License: nonfree and unredistributable
[00:28] <michaelni> durandal_1707, configure options
[00:28] <michaelni> ?
[00:29] <durandal_1707> libx264 is gpl and --enable-gpl is not specified
[00:29] <michaelni> so you add --enable-gpl
[00:30] <durandal_1707> wtf it can not found x264
[00:30] <burek> i mean how wrong is that, anybody can get the ffmpeg's source code from git and all the needed libs and compile whatever one wants, but the one cannot do all that and just leave the compiled result for the others to use it
[00:30] <durandal_1707> ah i get it, forgot to add additional flags ... linux centric configure script
[00:31] <burek> what is actually being protected in all that..?
[00:31] <burek> makes no sense to me
[00:32] <michaelni> burek, you can build a ffmpeg binary with all the gpl compatible things and distribute it with source under GPL
[00:32] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7227202 please review
[00:32] <durandal_1707> ahh, ignore me
[00:32] <burek> michaelni, ok, that was my goal in the first place
[00:32] <durandal_1707> i have enabled https protocol which links to tls/openssl
[00:32] <durandal_1707> that will make nonfree crap
[00:33] <burek> :)
[00:33] <burek> now, is there a list what is gpl and what's not, or should I just.. you know.. grep.. :)
[00:33] <durandal_1707> so you can not build with https protocol support
[00:33] <durandal_1707> s/build/distribute
[00:34] <michaelni> a  grep nonfree configure shows
[00:34] <michaelni> libaacplus, libfaac and openssl
[00:34] <durandal_1707> but note if you distribut binary of x264 you can get sued
[00:34] <durandal_1707> same apply to lame
[00:35] <michaelni> same apply if you disztribute anything
[00:35] <durandal_1707> but probably same applies to whole FFmpeg :))
[00:35] <burek> the only obvious result of all that crap is just to make it harder for people to do it themselves
[00:35] <burek> because, frankly, anyone can do it, in the end, compile whatever he needs, and use it
[00:35] <michaelni> durandal_1707, you are spreading FUD ....
[00:36] <burek> but if you do that, to make it easier for them, then no-no :)
[00:36] <michaelni> ffmpeg did not receice a single letter or email from anyone about h264, mp3 or anything in that class
[00:36] <michaelni> ever
[00:37] <durandal_1707> michaelni: FUD spreads like virus
[00:37] <michaelni> i know
[00:37] <michaelni> people that distribute commercial apps that use h264 decoders like ffmpeg have gotten polite mails asking them to pay some patent royalities
[00:38] <michaelni> iam not aware of anyone being sued by anyone
[00:39] <michaelni> burek, in what contry are you btw?
[00:39] <michaelni> patents are not a global thing
[00:39] <burek> wild west..
[00:39] <burek> called serbia
[00:40] <burek> i think that even the end of the world of 2012 will come in serbia at 2062
[00:41] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7227202 please review
[00:42] <burek> i mean i just wanted to compile ffmpeg that anyone can compile itself anyway.. just to make it easier for people who don't want to dive into compiling and stuff..
[00:42] <burek> there is no benefit for myself in that, that's why i think it's nonsene
[00:42] <burek> nonsense*
[00:43] <durandal_1707> well some people distribute static mplayer
[00:43] <durandal_1707> and that have ffmpeg with it
[00:44] <durandal_1707> but i'm not sure what linux distros are distributing 
[00:44] <michaelni> OanaStratulat, ok confirmed excessive mem allocation
[00:44] <CIA-49> ffmpeg: 03Carl Eugen Hoyos 07master * rb6b0067511 10ffmpeg/configure: 
[00:44] <CIA-49> ffmpeg: Assume architecture arm for machine hardware name iPad*.
[00:44] <CIA-49> ffmpeg: uname -m returns for example "iPad2,1".
[00:44] <CIA-49> ffmpeg: 03Hendrik Leppkes 07master * rb9d8af0327 10ffmpeg/libavcodec/mlpdec.c: mlpdec: fix channel order for wide 7.1 truehd layouts
[00:44] <CIA-49> ffmpeg: 03Carl Eugen Hoyos 07master * r1af0ace3a4 10ffmpeg/libavcodec/mlpdec.c: 
[00:44] <CIA-49> ffmpeg: Fix channel order for some less common TrueHD layouts.
[00:44] <CIA-49> ffmpeg: Reported and tested by TDiTP_
[00:44] <CIA-49> ffmpeg: Fixes ticket #779.
[00:44] <CIA-49> ffmpeg: 03Carl Eugen Hoyos 07master * r9a972b5111 10ffmpeg/libavcodec/acelp_vectors.c: Cosmetics: Fix indentation after last commit.
[00:44] <CIA-49> ffmpeg: 03Carl Eugen Hoyos 07master * r290e7eb77b 10ffmpeg/libavcodec/acelp_vectors.c: 
[00:44] <CIA-49> ffmpeg: Fix possible endless loop when decoding amr.
[00:44] <CIA-49> ffmpeg: Fixes bug 151.
[00:44] <CIA-49> ffmpeg: Reviewed-by: Vitor Sessak
[00:46] <michaelni> burek, durandal_1707: http://packages.ubuntu.com/search?keywords=x264
[00:46] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7226211 please accept
[00:46] <michaelni> so if someone gets "sued" they will go after ubuntu first
[00:46] <michaelni> and that hasnt happened
[00:47] <michaelni> also why will they go after ubuntu instead of a guy from serbia because people sue to get money
[00:47] <michaelni> so one sues the one that has most money
[00:48] <durandal_1707> http://www.freshports.org/audio/lame/
[00:48] <durandal_1707> package is not available
[00:49] <burek> but, the ironic thing is that i don't make any profit out of that
[00:49] <burek> so, it shouldn't be seen as a patent infrigement
[00:49] <michaelni> burek, what patent ? is there a valid patent in serbia on h264 at all ?
[00:50] <michaelni> or mp3 ...
[00:50] <burek> are you kidding me? :)
[00:50] <burek> we don't even have a law about the internet yet :)))
[00:51] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7226211 please review
[00:51] <burek> anyway, by building a publicly available static gpl-ed only ffmpeg, there would be no problems, right?
[00:51] <burek> legal problems*
[00:52] <burek> that's what linux distros are doing, just they use shared libs not static
[00:52] <durandal_1707> ask your lawyer
[00:52] <durandal_1707> :))
[00:52] <burek> :)
[00:52] <burek> I was always wondering why is there no one big static binary to download for linux :)
[00:53] <burek> always some dependencies hell or a manual compilation
[00:53] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7227203 please accept
[00:53] <durandal_1707> static or shared is irrelevant for this patents crap
[00:54] <durandal_1707> thay can not sue you for distributing source (that is my understanding)
[00:55] <durandal_1707> but dunno about libdvdcss and similar
[00:56] <michaelni> durandal_1707, sure they can, you can get sued for anything, just look in the news what weird things people sue over
[00:56] <ohsix> you are mixing the license problem and the patent problem, if you put something together from licenses that you invalidate, and it says you can't distribut ethe result; then you can't
[00:57] <durandal_1707> ohsix: yes we already resolved that
[00:57] <ohsix> i just read the buffer and that's not what it looked like
[00:59] <durandal_1707> ffmpeg can not be distributed if compiled with openssl/libfaac/libaacplus
[01:00] <durandal_1707> that is what configure gives
[01:00] <Compn> take this license talk to #flood :P
[01:00] Action: Compn runs back afk
[01:00] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7227203  please review
[01:03] <burek> so, can i, or can i not, build a one static ffmpeg binary, consisted of all the GPL libraries (encoders, formats..)
[01:03] <burek> i mean am i allowed to
[01:04] <durandal_1707> yes you are allowed to build
[01:04] <Compn> you can do whatever you want, if you arent going to distribute the ffmpeg binary :P
[01:05] <durandal_1707> just make sure that if you want to distribute it that configure output does not give output with nonfree word
[01:05] <ohsix> most of the termination clauses trigger only if you distribute the combined thing
[01:08] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7227203 please review
[01:09] <durandal_1707> michaelni: smjpeg actually have sounds too, and mplayer have demuxer for it
[01:12] <michaelni> OanaStratulat, segfault confirmed
[01:13] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7226210 please accept
[01:13] <michaelni> durandal_1707, can you port the demuxer to ffmpeg ? :)
[01:13] <durandal_1707> ohh some even made issue for this
[01:13] <durandal_1707> michaelni: should be straighforward
[01:20] <OanaStratulat> michaelni: http://www.google-melange.com/gci/task/view/google/gci2011/7226210  please review
[01:23] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7224213 please accept
[01:27] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7224213 please review
[01:27] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r3bf1d787b5 10ffmpeg/libavcodec/adx_parser.c: 
[01:27] <CIA-49> ffmpeg: adx_parser: Fix infinite loop.
[01:27] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:27] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * rbdd62a615a 10ffmpeg/libavcodec/adxdec.c: 
[01:27] <CIA-49> ffmpeg: adxdec: Do not require extradata.
[01:27] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:27] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r7859740c6a 10ffmpeg/libavcodec/adx_parser.c: 
[01:27] <CIA-49> ffmpeg: adx_parser: rewrite.
[01:27] <CIA-49> ffmpeg: The previous code ended in multiple different infinite
[01:27] <CIA-49> ffmpeg: loops. See stl_ten_1_big.sfd as example with and without zzuf
[01:27] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:27] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r6b6b84ae16 10ffmpeg/libavcodec/adxdec.c: 
[01:27] <CIA-49> ffmpeg: adxdec: Fix division by zero
[01:27] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:31] <OanaStratulat> has the src of ffmpeg been updated?
[01:38] <Compn> its contstantly updated
[01:38] <Compn> you have to constantly do git pull
[01:38] <OanaStratulat> michaelni: is the bug ok/
[01:40] <burek> no its not :)
[01:40] <burek> it needs to be debuged :))
[01:44] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * raf3f2a87da 10ffmpeg/libavformat/mov.c: 
[01:44] <CIA-49> ffmpeg: mov: detect EOF in mov_read_dref()
[01:44] <CIA-49> ffmpeg: This fixes a (near) infinite loop and Ticket783
[01:44] <CIA-49> ffmpeg: Bug found by: cosminamironesei
[01:44] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:53] <OanaStratulat> michaelni: do you have some time to review http://google-melange.appspot.com/gci/task/view/google/gci2011/7224213 ?
[01:56] <michaelni> OanaStratulat, iam quite busy atm, maybe someone else can help, burek can also do things on gci i think
[01:56] <OanaStratulat> burek: can you review tasks?
[01:57] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r1c73391d7d 10ffmpeg/libavformat/siff.c: 
[01:57] <CIA-49> ffmpeg: siff: Fix excessive memory allocation.
[01:57] <CIA-49> ffmpeg: Bug found by: Oana Stratulat
[01:57] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:58] <cbsrobot> I was intrigued by the logo competition ... maybe I should not have started working on it
[01:59] <cbsrobot> http://bayimg.com/palHoAaDh
[01:59] <CIA-49> ffmpeg: 03Tomas Härdin 07master * r31029e30fa 10ffmpeg/libavformat/mxfdec.c: 
[01:59] <CIA-49> ffmpeg: mxfdec: Move mxf_read_packet*() near the bottom of the file
[01:59] <CIA-49> ffmpeg: This simplifies future commits.
[01:59] <CIA-49> ffmpeg: 03Tomas Härdin 07master * r2eba8da675 10ffmpeg/ (3 files in 2 dirs): (log message trimmed)
[01:59] <CIA-49> ffmpeg: mxfdec: Hybrid demuxing/seeking solution
[01:59] <CIA-49> ffmpeg: This uses the old demuxing code for OP1a and separate demuxing code for OPAtom.
[01:59] <CIA-49> ffmpeg: Timestamp output is added to the old demuxing code.
[01:59] <CIA-49> ffmpeg: The seeking code is made to seek to the start of the desired EditUnit only,
[01:59] <CIA-49> ffmpeg: from which the normal demuxing code takes over (if OP1a). This means we don't
[01:59] <CIA-49> ffmpeg: use delta entries or slices, only StreamOffsets.
[01:59] <CIA-49> ffmpeg: 03Tomas Härdin 07master * rf98726551c 10ffmpeg/libavformat/mxfdec.c: mxfdec: Remove dead/useless code
[01:59] <CIA-49> ffmpeg: 03Tomas Härdin 07master * r36c305f9db 10ffmpeg/libavformat/mxfdec.c: 
[01:59] <CIA-49> ffmpeg: mxfdec: Add mxf_edit_unit_absolute_offset()
[01:59] <CIA-49> ffmpeg: This maps an EditUnit in an index table to the corresponding absolute offset
[01:59] <CIA-49> ffmpeg: 03Tomas Härdin 07master * r8ce41721a4 10ffmpeg/libavformat/mxfdec.c: 
[01:59] <CIA-49> ffmpeg: mxfdec: Add MXFIndexTables
[01:59] <CIA-49> ffmpeg: This structs collects MXFIndexTableSegments belonging to one IndexSID and some
[01:59] <CIA-49> ffmpeg: other bits of useful information.
[01:59] <CIA-49> ffmpeg: 03Tomas Härdin 07master * rb56adc7842 10ffmpeg/libavformat/mxfdec.c: 
[01:59] <CIA-49> ffmpeg: mxfdec: Replace zero IndexDurations with st->duration
[01:59] <CIA-49> ffmpeg: The mxf_d10 muxer outputs files with zero IndexDurations..
[01:59] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * ra9ef21bbe2 10ffmpeg/: (log message trimmed)
[01:59] <CIA-49> ffmpeg: Merge remote-tracking branch 'tjoppen/proper_mxf_track_linking'
[01:59] <CIA-49> ffmpeg: * tjoppen/proper_mxf_track_linking:
[01:59] <CIA-49> ffmpeg:  mxfdec: Don't parse slices or DeltaEntryArrays
[02:00] <CIA-49> ffmpeg:  mxfdec: Remove dead/useless code
[02:00] <CIA-49> ffmpeg:  mxfdec: Hybrid demuxing/seeking solution
[02:00] <CIA-49> ffmpeg:  mxfdec: Add mxf_edit_unit_absolute_offset()
[02:00] <CIA-49> ffmpeg: 03Tomas Härdin 07master * r0e764cf541 10ffmpeg/libavformat/mxfdec.c: 
[02:00] <CIA-49> ffmpeg: mxfdec: Don't parse slices or DeltaEntryArrays
[02:00] <CIA-49> ffmpeg: The most recent demuxing/seeking code doesn't need them
[02:00] <CIA-49> ffmpeg: 03Tomas Härdin 07master * r2212e040b7 10ffmpeg/ (libavformat/mxfdec.c tests/ref/seek/lavf_mxf): 
[02:00] <CIA-49> ffmpeg: mxfdec: Add "fake" index to MXFIndexTable to assist seeking
[02:00] <CIA-49> ffmpeg: This changes mxf_compute_ptses() to be used for MXFIndexTable, and also adds
[02:00] <CIA-49> ffmpeg: code for computing the fake index to it.
[02:00] <CIA-49> (1 lines omitted)
[02:16] <michaelni> OanaStratulat, reproduced crash
[02:17] <OanaStratulat> michaelni: http://google-melange.appspot.com/gci/task/view/google/gci2011/7196354 please accept
[02:18] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * rf371396dfb 10ffmpeg/libavformat/rl2.c: 
[02:18] <CIA-49> ffmpeg: rl2demux: Fix FPE
[02:18] <CIA-49> ffmpeg: Fixes Ticket788
[02:18] <CIA-49> ffmpeg: Bug found by Oana Stratulat
[02:18] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:22] <OanaStratulat> michaelni: please accept just to submit my work and then i go to sleep because  have school 2morrow and it`s kinda late here in Romania
[02:27] <michaelni> OanaStratulat, done, good night :)
[02:27] <OanaStratulat> thx, same 2 you
[02:28] <OanaStratulat> will be more tasks with finding bugs submitted?
[02:29] <michaelni> are there none left ?
[02:29] <OanaStratulat> i think 10 or 9
[02:29] <michaelni> hmm, thats not much
[02:29] <OanaStratulat> but if i stay and search for bugs i think i will finish them 
[02:30] <michaelni> iam not sure if i can add more tasks now if iam already past googles deadline for that
[02:30] <michaelni> i think there where only 2 dates when new tasks would become vissible
[02:30] <OanaStratulat> ok. good night 
[02:31] <michaelni> good night
[02:31] <OanaStratulat> btw the task is marked as completed 
[02:31] <OanaStratulat> http://google-melange.appspot.com/gci/task/view/google/gci2011/7196354
[02:31] <michaelni> ill look when i have time
[02:55] <caytchen> so, whats the expected behavior for read/write handlers passed to avio_alloc_context? is not reading buf_size bytes completely fine?
[03:18] <CIA-49> ffmpeg: 03Anton Khirnov 07master * r08f7af0c37 10ffmpeg/libavformat/ (Makefile oma.c): lavf: rename oma.c -> omadec.c
[03:18] <CIA-49> ffmpeg: 03Uoti Urpala 07master * r488a5b3fdd 10ffmpeg/libavcodec/tmv.c: 
[03:18] <CIA-49> ffmpeg: tmv decoder: set correct pix_fmt
[03:18] <CIA-49> ffmpeg: Previously the decoder only worked if the user had set avctx->pix_fmt
[03:18] <CIA-49> ffmpeg: manually. For some reason the libavformat tmv demuxer sets this, so
[03:18] <CIA-49> ffmpeg: the problem was not visible in avplay etc.
[03:18] <CIA-49> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje at gmail.com>
[03:18] <CIA-49> ffmpeg: 03Michael Karcher 07master * rf40e7eb573 10ffmpeg/ (6 files in 3 dirs): 
[03:18] <CIA-49> ffmpeg: lavf: add OpenMG audio muxer.
[03:18] <CIA-49> ffmpeg: Signed-off-by: Anton Khirnov <anton at khirnov.net>
[03:18] <CIA-49> ffmpeg: 03Anton Khirnov 07master * ree20b332c8 10ffmpeg/libavformat/ (Makefile oma.c oma.h omadec.c): omadec: split data that will be used in the muxer to a separate file.
[03:18] <CIA-49> ffmpeg: 03Ronald S. Bultje 07master * rd49352c7cc 10ffmpeg/libswscale/ (swscale.c utils.c): (log message trimmed)
[03:18] <CIA-49> ffmpeg: swscale: fix overflows in vertical scaling at top/bottom edges.
[03:18] <CIA-49> ffmpeg: This fixes integer multiplication overflows in RGB48 output
[03:18] <CIA-49> ffmpeg: (vertical) scaling as detected by IOC. What happens is that for
[03:18] <CIA-49> ffmpeg: thread count.
[03:18] <CIA-49> ffmpeg: 03Justin Ruggles 07master * rbc78ceec2b 10ffmpeg/libavcodec/4xm.c: 
[03:18] <CIA-49> ffmpeg: 4xm: remove unneeded check for remaining unused data.
[03:18] <CIA-49> ffmpeg: This is unnecessary complication that only prints a message and does not
[03:18] <CIA-49> ffmpeg: affect decoding.
[03:19] <CIA-49> ffmpeg: 03Mans Rullgard 07master * r2a216ca2ef 10ffmpeg/doc/APIchanges: 
[03:19] <CIA-49> ffmpeg: APIchanges: fill in revision for AVFrame.age deprecation
[03:19] <CIA-49> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[03:19] <CIA-49> ffmpeg: 03Mans Rullgard 07master * r8400b126ac 10ffmpeg/ (9 files in 3 dirs): 
[03:19] <CIA-49> ffmpeg: avcodec: deprecate AVFrame.age
[03:19] <CIA-49> ffmpeg: This was intended as an optimisation for skipped blocks in MPEG2
[03:19] <CIA-49> ffmpeg: P-frames and never used elsewhere. Removing this "optimisation"
[03:19] <CIA-49> ffmpeg: speeds up MPEG2 decoding by 1-2% (ARM Cortex-A9).
[03:19] <CIA-49> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[03:45] <durandal_1707> is there a way to set duration of file in msec?
[03:48] <durandal_1707> can i use duration from AVFormatContext?
[03:50] <michaelni> what exactly do you want to do ?
[03:51] <durandal_1707> format of smjpeg in header have duration of file in mili seconds
[04:16] <caytchen> if I use custom IO for AVFormatContext, can I leave the filename as NULL for avformat_open_input?
[04:17] <michaelni> durandal_1707, sorry for my slow awnser, yes you can set duration, its in AV_TIME_BASE units
[04:18] <michaelni> *AV_TIME_BASE/1000
[04:18] <durandal_1707> michaelni: no problem there are so many demuxers so i just explore
[04:19] <durandal_1707> there is even source code for smjpeg encoder (muxer really)
[04:20] <michaelni> caytchen, filename is an array so you cant set it ti NULL but you can put a zero lenbth string in there
[04:22] <caytchen> well the question really is, will it completely disregard filename if theres an IO context or does it have functions beyond that?
[04:24] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r29179239f5 10ffmpeg/libavcodec/proresdec2.c: 
[04:24] <CIA-49> ffmpeg: proresdec: add UPDATE_CACHE, should fix decoding with -Os
[04:24] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:24] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * rf85334f58e 10ffmpeg/libavcodec/eatqi.c: 
[04:24] <CIA-49> ffmpeg: tqi: Fix passing errors from the MB decode. This silences some valgrind warnings.
[04:24] <CIA-49> ffmpeg: Fixes second half of Ticket794
[04:24] <CIA-49> ffmpeg: Bug found by: Oana Stratulat
[04:24] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:25] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r608708009f 10ffmpeg/libavcodec/adpcm.c: 
[04:25] <CIA-49> ffmpeg: adpcm: Fix crash
[04:25] <CIA-49> ffmpeg: Fixes half of Ticket794
[04:25] <CIA-49> ffmpeg: Bug found by: Oana Stratulat
[04:25] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:25] <michaelni> caytchen, my guess is something will access the filename no matter what io context is used
[04:47] <durandal_1707> how to make read_packet to not read what read_header already did?
[04:48] <durandal_1707> i could use ugly hack but i do not like that
[09:33] <OanaStra_> http://www.google-melange.com/gci/task/view/google/gci2011/7196354   Please review
[11:12] <OanaStratulat> michaelni: http://www.google-melange.com/gci/task/view/google/gci2011/7196354 please review this task, use the corruptfile2 and valgrind2
[12:07] <michaelni> OanaStratulat, i can reproduce it but it might be the same issue as Ticket792, iam not sure, i need to investigate this first to be sure if or if not they are the same
[12:23] <kralikba> hello
[12:23] <kralikba> I'm trying to compile ffmpeg on arm but I'm getting errors of an undefined reference to log2f and exp2f at link-time 
[12:23] <kralikba> LD ffmpeg_g
[12:25] <av500> -lm?
[13:55] <OanaStratulat> michaelni: some luck?
[13:58] <michaelni> OanaStratulat, i was busy :( had no time yet  to look
[13:58] <OanaStratulat> michaelni: ok no prob, i`m waiting
[14:13] <OanaStratulat> michaelni: can you help me to set up lcov on my system because i want to take some hard tasks from GCI to add some files to the fate system, and i`m a bit confused how to install that software
[14:15] <ubitux> what system are you running?
[14:16] <j-b> Win 3.11
[14:17] <ubitux> you have to install the lcov package, compile ffmpeg with --enable-coverage (--enable-gpl would be nice too), run fate (make fate SAMPLES=...), and then make coverage-html
[14:18] <ubitux> install lcov depends on your install
[14:18] <ubitux> installing*
[14:18] <ubitux> your system*
[14:18] <ubitux> .
[14:25] <OanaStratulat> 10x
[14:28] <OanaStratulat> "compile ffmpeg with --enable-coverage" = ./configure --enable-coverage? 
[14:28] <ubitux> yes
[15:00] <OanaStratulat> michaelni: had time?
[15:04] <OanaStratulat> ubitux: when i try to run lcov it says multiple times "cannot open source file" have any idea?
[15:05] <ubitux> why would you run lcov directly?
[15:06] <OanaStratulat> i don`t get it, be more explicit
[15:07] <ubitux> you say you want to run lcov, but you don't need to
[15:07] <ubitux> this is handled by the ffmpeg with "make coverage-html"
[15:07] <cbreak-work> if I have a small patch for ffmpeg, is it prefered if I send it to a mailing list, or provide a git repository with the commit?
[15:07] <ubitux> (of course, after you did a first run)
[15:08] <ubitux> cbreak-work: mailing list is fine
[15:08] <OanaStratulat> when i try to enter make coverage-html it says there is no make for it
[15:09] <ubitux> OanaStratulat: do you use the latest git?
[15:09] <OanaStratulat> ubitux: the git from the download site ffmpeg
[15:10] <OanaStratulat> ubitux: isn`t that the latest?
[15:10] <ubitux> yes it is
[15:10] <ubitux> the coverage-html should exists then
[15:10] <ubitux> do you have matches with "git grep coverage-html" ?
[15:10] <OanaStratulat> ubitux: i`ll try to clone another time, maybe i`m did something wierd
[15:14] <OanaStratulat> in the makefile i have coverage-html
[15:14] <OanaStratulat> i`m compiling angain
[15:14] <OanaStratulat> again
[15:17] <ubitux> check if --enable-coverage was effective
[15:17] <ubitux> (you should have some special cflags)
[15:25] <OanaStratulat> when i make fate it gives an error http://pastebin.com/yy2KWmeR
[15:26] <ubitux> mmh did you sync your samples recently? (make fate-rsync SAMPLES=...)
[15:28] <cbreak-work> It's basically this: http://paste.the-color-black.net/216225, I'll try to send an email with the patch to the mailing list, as soon as I manage to subscribe :)
[15:36] <OanaStratulat> ubitux: yes, the same eror
[15:36] <OanaStratulat> error even if i sync the samples
[15:39] <OanaStratulat> ubitux: any other ideas?
[15:41] <ubitux> it's surprising since this failure doesn't appear in fate
[15:41] <ubitux> sorry i'm helpless atm
[15:42] <OanaStratulat> ok no prob
[15:43] <michaelni> OanaStratulat, missing zlib maybe 
[15:44] Action: michaelni goes back to investigating that bug
[15:44] <OanaStratulat> idea how to fix that because i don`t have any clue what zlib is
[15:46] <OanaStratulat> michaelni: had any time to look at my bug?
[15:51] <michaelni> iam currently :)
[15:52] <OanaStratulat> i get the same error from the fate system on 2 different machines
[15:53] <michaelni> ubuntu ?
[15:54] <OanaStratulat> yes
[15:54] <OanaStratulat> 2 ubuntu machines
[15:54] <michaelni> apt-get install zlib1g-dev
[15:55] <OanaStratulat> testing now
[15:55] <OanaStratulat> the same
[15:56] <michaelni> OanaStratulat, you did run ./configure ?
[15:56] <michaelni> after installing zlib-dev ...
[15:56] <OanaStratulat> doing now
[15:57] <ubitux> we should think of a fate with --disable-zlib
[15:57] <michaelni> yes
[15:57] <michaelni> also theres mans zlib reimplemantation but i had out of array writes with it ...
[15:58] <ubitux> my box is painfully handling the valgrind test so i won't be able to propose one until a while :p
[15:58] <ubitux> when i'm rich i may do something about this :)
[15:59] <michaelni> ubitux, you could run the valgrind test just 1 or 2 times a day
[15:59] <michaelni> if the load is a problem
[15:59] <ubitux> yes sure
[16:00] <ubitux> atm it is only used for this, i guess i could queue a --disable-zlib box
[16:00] <ubitux> i'll think about it :)
[16:00] <michaelni> also think about a address sanitizer box and a IOC box (both these are alot faster than valgrind)
[16:01] <michaelni> and more usefull than --disable-zlib
[16:03] <ubitux> mmh yep ok
[16:05] <ubitux> i wonder if there are tests with dependencies; only 667/990 tests are run with valgrind
[16:08] <michaelni> ubitux, seektests depend on the tests that build files
[16:08] <ubitux> mmh, ok
[16:12] <OanaStratulat> http://pastebin.com/gMaxuX6d  this is my error now?
[16:12] <OanaStratulat> without "?"
[16:24] <OanaStratulat> michaelni: http://www.google-melange.com/gci/task/view/google/gci2011/7211212  please accept
[16:38] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r9c5104e84e 10ffmpeg/libavformat/dsicin.c: 
[16:38] <CIA-49> ffmpeg: dsicin demuxer: Fix excessive malloc()
[16:38] <CIA-49> ffmpeg: use ffio_limit()
[16:38] <CIA-49> ffmpeg: Fixes Ticket 790
[16:38] <CIA-49> ffmpeg: Bug found by: Oana Stratulat
[16:38] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:38] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r90cad5edf6 10ffmpeg/libswscale/utils.c: 
[16:38] <CIA-49> ffmpeg: swscale: Fix several (real world) integer overflows.
[16:38] <CIA-49> ffmpeg: Fixes Ticket795
[16:38] <CIA-49> ffmpeg: Bug Found by: Oana Stratulat
[16:38] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:38] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * re39eeb10d1 10ffmpeg/libavformat/ (avio_internal.h utils.c): 
[16:38] <CIA-49> ffmpeg: lavf: split out ffio_limit()
[16:38] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:38] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r32a92d3258 10ffmpeg/libswscale/utils.c: 
[16:38] <CIA-49> ffmpeg: sws: Print an error in case of av_malloc() failure
[16:38] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:38] <CIA-49> ffmpeg: 03Michael Niedermayer 07master * r03a9c9932d 10ffmpeg/libavcodec/eamad.c: 
[16:38] <CIA-49> ffmpeg: eamad: fix excessive memory allocation.
[16:38] <CIA-49> ffmpeg: Fixes Ticket792
[16:38] <CIA-49> ffmpeg: Bug found by Oana Stratulat
[16:38] <CIA-49> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:41] <ubitux> oh CIA-49 is alive again
[17:21] <michaelni> our get ISO specs GCI task reached its deadline
[17:22] <michaelni> ... without reply from anyone from ISO
[17:22] <michaelni> http://www.google-melange.com/gci/task/view/google/gci2011/7185251
[17:22] <michaelni> what should i do ?
[17:22] <av500> what a surprise
[17:22] <michaelni> well no surprise :)
[17:22] <michaelni> it was an inpossible task
[17:23] <av500> task it to Tom Cruise then
[17:23] <michaelni> the student spend time researching whom to contact and contacted everyone
[17:24] <michaelni> iam not happy to unassign / fail it :(
[17:24] <av500> well, you knew the outcome before
[17:24] <av500> but then, so did he
[17:24] <av500> as did ISO
[17:27] <ubitux> michaelni: "request should call the sources request in a loop until a frame is output from thumbnail" -> isn't request_frame() (in thumbnail) supposed to be called only after poll_frame() says 1 frame can be output for thumbnail? so we basically have to return a negative value in case it is called when it couldn't?
[17:27] <ubitux> instead of a loop
[17:31] <michaelni> ubitux, how should that work, if the application needs a frame and poll says 0
[17:31] <michaelni> printf("shit we are stuck");    ;)
[17:32] <ubitux> but what's the point of poll_frame() if we can actually force a request_frame() and get one?
[17:34] <michaelni> if you have 2 outputs from where do you request ? 
[17:34] <michaelni> picking the wrong too often may lead to huge buffer requirements
[17:36] <ubitux> mmh ok
[17:36] <ubitux> do we have use case for this already?
[17:36] <kierank> there aren't that many iso standards which we don't have
[17:36] <kierank> it's a pointless exercise
[17:36] <ubitux> we have access to iso standards?
[17:37] <ubitux> (legally speaking)
[17:37] <kierank> what do you want and i'll get it
[17:37] <kierank> ok some i have are a tad dubious
[17:37] <ubitux> i personally have access to most of them but well&
[17:38] <ubitux> i wonder if i can even quote them on the mailing list without getting into legal troubles :p
[17:38] <kierank> pffft they don't give a shit
[17:39] <ubitux> oh sure :)
[17:39] <kierank> they make a lot of money from companies buying them so they're not going to give them out for free
[17:39] <ubitux> that's just an ethical question i'd say :p
[17:47] <Compn> we should ask our google friends if they can share iso with us
[17:51] <ubitux> michaelni: also a strange thing: in poll_frame() i try to poll_frame() from the inlink, and if at least one frame is available and i'm sure the next time end_frame() is fed with a frame it will output a thumb, then i return 1; but in that case end_frame() is not called
[17:51] <caytchen> you could also get a university library card and download the ones you need for private use ;) but I guess you already have that.
[17:51] <ubitux> request_frame() is called though
[17:52] <ubitux> so should i manually call for end_frame() through request_frame()?
[17:52] <ubitux> that sounds wrong
[17:53] <ubitux> request_frame() from yadif seems to actually do that (with return_frame()) though
[17:54] <ubitux> caytchen: i was more concerned about open discussion about the standard, or easy access to all the developers
[17:58] <michaelni> ubitux, if you "have" a frame then you should be able to output it from request_frame
[17:59] <michaelni> if you dont you call your sources request
[17:59] <michaelni> which in turn at some point somehow will call your start/end frame with a new frame
[17:59] <ubitux> i don't call request_frame() when i have one, but only when input link anounced it has the one i need to return one
[18:02] <ubitux> < michaelni> if you dont you call your sources request // you mean call avfilter_request_frame() on the inlink and this will trigger the start/end frame with the new "announced" frame?
[18:02] <ubitux> (in request_frame())
[18:03] <michaelni> sounds correct
[18:04] <ubitux> it seems to do the trick, ok
[18:04] <ubitux> but then, wouldn't it be more simple to just announced i have one thumb in poll when i actually do have one?
[18:05] <ubitux> announce*
[18:11] <cbsrobot> so no comments on my logo proposition? http://bayimg.com/palHoAaDh
[18:13] <ubitux> heh nice one
[18:13] <ubitux> i'm not sure the pixel art style is really in the spirit of winter
[18:13] <ubitux> it would have do the trick with the blue (it reminds ice)
[18:14] <ubitux> but with the white, it looks more like a new minecraft texture for snow
[18:14] <cbsrobot> colors are a bit random
[18:14] <cbsrobot> it's more about the idea ...
[18:16] <Tjoppen> >site can't show an image without javascript
[18:16] <Tjoppen> amazing
[18:17] <Tjoppen> as for the logo, it seems ok?
[18:26] <michaelni> ubitux, cbsrobot we have a very beautifull winter logo already
[18:26] <michaelni> http://www.google-melange.com/gci/task/view/google/gci2011/7191489
[18:30] Action: michaelni fires up gimp to downsize it and to send it to the ML
[18:30] <michaelni> now what was the size of our current logo ...
[18:34] <gnafu> Ooh, I like that.  Festive, but not overdone.
[18:35] <gnafu> Or perhaps festive isn't the right word, but anyway...
[19:06] <ubitux> michaelni: ah indeed, and nice one
[20:17] <pasteeater> michaelni: are there logo gci tasks, other than the winter one?
[20:18] <pasteeater> i registered as mentor a few days ago and never heard back, but that was good because i had to leave unexpectedly for a few days
[20:24] <michaelni> pasteeater, we are mostly done with the logos sadly
[20:25] <michaelni> pasteeater, http://www.google-melange.com/gci/task/view/google/gci2011/7204205
[20:25] <michaelni> only one left
[20:26] <michaelni> but you can help with anything else thats left
[20:26] <michaelni> assigning tasks when a student wants to work on it
[20:26] <michaelni> awnsering questions
[20:26] <michaelni> ...
[20:29] <pasteeater> you can add holiday logos if you want more.
[20:32] <ubitux> michaelni: what do you plan to do about the ABI issue in AVERROR_BUG?
[20:32] <ubitux> (they just did the same as Nicolas' patch introducing AVERROR_BUG, but breaking ABI)
[20:33] <ubitux> (poorly done btw, break alphabetical order & no doc/APIChanges update)
[20:33] <michaelni> pasteeater, iam notz sure if we can add more tasks
[20:33] <ubitux> (and no bump)
[20:36] <michaelni> ubitux, dunno yet, guess just document 
[20:36] <michaelni> its not really important
[20:37] <michaelni> noone checks error codes except EOF/EAGAIN
[20:41] <ubitux> ok
[21:10] <michaelni> pasteeater, you can try to add more logo tasks if you want
[21:10] <michaelni> but iam not sure if they will show up
[21:18] <caytchen> is avcodec_decode_audio2 the actual core api for decoding audio or just some highlevel convienence function? it seems limited to 16 bit samples
[21:19] <nevcairiel> its decode_audio4 these days
[21:19] <nevcairiel> although last stable probably still was 3
[21:20] <caytchen> ah, okay, heh. its hard to find any documentation for the general design of the pipeline
[21:22] <Daemon404> michaelni, the downsize logo could use some AA :P
[21:22] <Daemon404> s/downsize/downsized/
[21:27] <Daemon404> tho i guess the lines are indeed fairly thing anyway
[21:27] Action: Daemon404 shrugs
[21:31] <michaelni> Daemon404, if you can downscale it better, do it
[21:31] <michaelni> ive used gimp
[21:31] <michaelni> and i agree the result is not that hot
[21:31] <Daemon404> nah, photoshops doesnt look very nice either
[21:32] <Daemon404> i wonder how inkspace would render the svg though
[21:34] <pasteeater> didn't notice the border until now. black/gray lines on all sides but top
[21:42] <pasteeater> slightly less jagged with complete border and smaller palette: http://akrobotics.com/FFmpeg_winter_smaller.png
[22:49] <durandal_1707> omg i dossed my own machine with ffplay
[22:51] <michaelni> durandal_1707, ?
[22:52] <durandal_1707> some stupid FreeBSD bug
[22:53] <michaelni> detected patent encumbered mp3 decoder, halting system ?
[22:54] <durandal_1707> noo, I'm using file as swap
[22:55] <durandal_1707> actually it was my fault, i put loop into smjpeg demuxer and it caused unlimited seeks, which stalled machine
[22:56] <caytchen> ´
[22:56] <durandal_1707> something scheduling related is buggy
[22:57] <durandal_1707> well considering i'm using CURRENT that is expected
[22:59] <durandal_1707> but now when I'm staring at code, i wonder how that happended because eof_reached should handle such situation
[23:03] <michaelni> pasteeater, are you here ?
[23:03] <michaelni> http://www.google-melange.com/gci/task/view/google/gci2011/7204205
[23:03] <michaelni> thats the last logo
[23:03] <michaelni> and the final one looks ok to me
[23:09] <pasteeater> michaelni: i see the butterfly in google image search.
[23:10] <michaelni> pasteeater, copyright problem ?
[23:11] <pasteeater> i'm not sure if it's subject to copyright.
[23:13] <michaelni> well you wanted to mentor the logos...
[23:14] <michaelni> thats your chance to find out if its ok or not, you can ask the student :)
[23:14] <michaelni> where its from
[23:15] <pasteeater> i'll ask
[23:15] <michaelni> thx :)
[23:15] <michaelni> iam already a bit exhausted by GCI ...
[23:19] <pasteeater> ...after this meeting i forgot about.
[23:19] <j-b> michaelni: told you so
[00:00] --- Tue Dec 20 2011