[Ffmpeg-devel-irc] ffmpeg-devel.log.20120125

[burek]

[00:02] <saste> ubitux: looks a timestamp issue
[00:02] <saste> they're chasing me lately
[00:03] <saste> ashowinfo should give some clues
[00:07] <ubitux> http://blackhole.ubitux.fr/ashowinfo  amovie=a.mp3,ashowinfo
[00:08] <ubitux> http://blackhole.ubitux.fr/ashowinfo2  amovie=a.mp3,aresample=8000,ashowinfo
[00:08] <ubitux> anything suspicious?
[00:09] <saste> uhm... tried here and works fine
[00:10] <iive> ubitux: just random question, does the -t work correctly without the resample ?
[00:10] <ubitux> iive: yes
[00:12] <iive> then it may be error in calculation of time in audio filter(s)
[00:12] <ubitux> maybe my mp3 is at fault it's strange
[00:13] <saste> ubitux: yes it's a timestamp issue, in the second output timestamps are weird
[00:13] <ubitux> maybe this file was generated when there was the mp3 header writing issue
[00:14] <saste> put an ashowinfo before and after the resample, it will show you how the timestamps are changed
[00:15] <ubitux> http://yozora-irc.net/~ubitux/resample-me-with-lavfi.mp3
[00:15] <ubitux> (if you want to try)
[00:16] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r2179b638e3 10ffmpeg/libavdevice/v4l.c: 
[00:16] <CIA-31> ffmpeg: v4l: fix compilation
[00:16] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:17] <ubitux> saste: http://pastie.org/3246498
[00:17] <ubitux> 'get this
[00:17] <ubitux> so what's exactly wrong here?
[00:18] <ubitux> pts are indeed changing but& how are they wrong?
[00:18] <saste> nothing obvious -> debugging time!
[00:19] <j-b> michaelni: someone kills our server :'(
[00:19] <j-b> or killed
[00:22] <michaelni> j-b, someone ?
[00:23] <j-b> or something
[00:23] <j-b> or an animal
[00:23] <j-b> or a star
[00:23] <j-b> or a galaxy
[00:23] <burek> SOPA :)
[00:28] <ubitux> a mushi.
[00:29] <iive> btw, it may be wise idea all projects to move to a .eu domains. or at least to avoid .com and .org
[00:30] <burek> :)
[00:31] <burek> moving domain won't help much if your server is in us or eu
[00:31] <burek> ive moved all of mine to asia
[00:35] <ubitux> behind the great wall?
[00:36] <burek> well, they still respect customers :)
[00:36] <ubitux> :)
[00:36] <burek> that's what i care about :)
[00:38] <ubitux> btw, i recommend the movie Bin Jip, like most of Ki-duk Kim movies
[00:51] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r749e5dc1fd 10ffmpeg/libavdevice/Makefile: 
[00:51] <CIA-31> ffmpeg: buildsystem: 10l add v4l entry back
[00:51] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:56] <Compn> michaelni : i'll ask BBB about libav's plans for avconv 
[00:57] <michaelni> Compn, do you think he wants to maintain it in ffmpeg ?
[00:57] <michaelni> ;)
[00:57] <Compn> lol
[00:58] <michaelni> btw, we still need that call for forgotten patches ....
[01:00] <kierank> so...who's going to fosdem?
[01:00] <michaelni> j-b, isnt there some hoster that you can mail / call and ask if he hears some animal noises from the server box?
[01:00] <Compn> michaelni : ok i'll send patch about 'forgotten patches' to news page...
[01:00] <Compn> animal noises ?
[01:00] <Compn> did i miss something? :)
[01:00] <michaelni> try streams.videolan.org
[01:01] <michaelni> or upload.ffmpeg.org
[01:01] <Compn> yeah, i noticed that the other day :\
[01:02] <michaelni> <j-b> or an animal
[01:05] <pasteeater> kierank: good luck on your talk.
[01:05] <j-b> michaelni: motherboard is likely dead
[01:06] <Compn> oh noes
[01:07] <Compn> michaelni : so is carl going to start up samples repo ?
[01:07] <Compn> and then mike will rsync from there ?
[01:07] <michaelni> i never got a reply from mike :(
[01:07] <Compn> well lets just do it that way, since we dont have access to libav server 
[01:08] <Compn> well maybe you guys are still dancing. just i'm bored ;)
[01:11] <michaelni> Compn, ill look into that and ill write mike another mail
[01:16] <Compn> bbb says he will be on vacation and he will talk to google after he gets back
[01:16] <Compn> and says elenril has plans for avconv , better ask him about that :P
[01:17] <michaelni> elenril talks with you ?
[01:17] <Compn> no
[01:17] <Compn> i havent tried contacting him
[01:18] <Compn> want for me to try ?
[01:18] <michaelni> there are some issues in his code that he should fix ...
[01:18] <michaelni> but he doesnt talk with me
[01:19] <Compn> if you have commits / patches for me to pass on, i can try to ask him about it
[01:19] <michaelni> no commits just bug reports
[01:19] <michaelni> its somewhere on the ML
[01:20] <michaelni> his command/option parsing code being insanely inefficient and slow for example
[01:20] Action: Daemon404 cant see that taking an axcessive amount of time
[01:21] <Daemon404> excessive*
[01:23] <michaelni> The problem is it takes O(n^4) or O(n^3) IIRC for n options, so its completely non scaleable
[03:51] <CIA-31> ffmpeg: 03Carl Eugen Hoyos 07master * r7988dd1b9a 10ffmpeg/libavcodec/libfaac.c: Fix multi-channel encoding with libfaac.
[06:52] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r5c011706bc 10ffmpeg/libavformat/nsvdec.c: 
[06:52] <CIA-31> ffmpeg: nsvdec: Fix use of uninitialized streams.
[06:52] <CIA-31> ffmpeg: Fixes CVE-2011-3940 (Out of bounds read resulting in out of bounds write)
[06:52] <CIA-31> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[06:52] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[06:52] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r59e95fa4a8 10ffmpeg/libavcodec/h263dec.c: 
[06:52] <CIA-31> ffmpeg: h263dec: Disallow width/height changing with frame threads.
[06:52] <CIA-31> ffmpeg: Fixes CVE-2011-3937
[06:52] <CIA-31> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[06:52] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[06:52] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r28d634711b 10ffmpeg/libavformat/avidec.c: (log message trimmed)
[06:52] <CIA-31> ffmpeg: avidec: Fix regression with chunks that are larger than the file.
[06:52] <CIA-31> ffmpeg: This commit makes the check specific to the case that needs it.
[06:52] <CIA-31> ffmpeg: Regression was introduced by
[06:52] <CIA-31> ffmpeg: commit 62adc60b97d854507d07a21b2f370ab5c69e6b7b
[06:52] <CIA-31> ffmpeg: Author: Michael Niedermayer <michaelni at gmx.at>
[06:53] <CIA-31> ffmpeg: Date: Fri Dec 16 06:13:04 2011 +0100
[06:53] <CIA-31> ffmpeg: 03Michael Niedermayer 07master * r668494acd8 10ffmpeg/ffmpeg.c: 
[06:53] <CIA-31> ffmpeg: ffmpeg: add image size check to codec_get_buffer()
[06:53] <CIA-31> ffmpeg: Fixes CVE-2011-3935
[06:53] <CIA-31> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[06:53] <CIA-31> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[14:02] <ubitux> why is there an auto inserted scale filter code in ffmpeg.c while lavfi can do it transparently already?
[14:03] <ubitux> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=ffmpeg.c;hb=HEAD#l645
[14:03] <Tjoppen> old code?
[14:03] <ubitux> possible
[14:03] <Tjoppen> "patch welcome"
[14:04] <ubitux> well that's easy to drop
[14:55] <ubitux> it seems to be required at some point
[15:05] <ubitux> the auto-insert scale filter seems to deal only with formats
[15:05] <ubitux> while the one in ffmpeg is for i/o codec width/height mismatch
[16:13] <CIA-31> ffmpeg: 03Nicolas George 07master * r4a68949cd8 10ffmpeg/libavfilter/Makefile: lavfi: Makefile: cosmetics: align FFLIBS.
[16:44] <CIA-108> ffmpeg: 03Michael Niedermayer 07master * rf9de136b17 10ffmpeg/libavformat/dv.c: 
[16:44] <CIA-108> ffmpeg: dv: check stype
[16:44] <CIA-108> ffmpeg: Fixes part1 of CVE-2011-3929
[16:44] <CIA-108> ffmpeg: Possibly fixes part of CVE-2011-3936
[16:44] <CIA-108> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[16:44] <CIA-108> ffmpeg: Reviewed-by: Roman Shaposhnik <roman at shaposhnik.org>
[16:44] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:44] <CIA-108> ffmpeg: 03Michael Niedermayer 07master * r5cb57a16ed 10ffmpeg/libavformat/dv.c: 
[16:44] <CIA-108> ffmpeg: dv: Fix null pointer dereference due to ach=0
[16:44] <CIA-108> ffmpeg: Fixes part2 of CVE-2011-3929
[16:44] <CIA-108> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[16:44] <CIA-108> ffmpeg: Reviewed-by: Roman Shaposhnik <roman at shaposhnik.org>
[16:44] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:08] <ramiro> hi, is alexander strasser around?
[17:09] Action: funman spots ramiro
[17:10] <funman> dunno who he is sorry
[17:14] <ubitux> ramiro: i don't think so; his nick is beastd iirc
[17:14] <iive> isn't that beastd?
[17:14] <burek> beastd
[17:14] <burek> :D
[17:14] <ubitux> what a star :)
[17:14] <burek> I had to search in logs, and as soon as I found it, I looked in here and both of you already had that nick written :D
[17:15] <burek> such a fail :)
[17:15] Action: gnafu chimes in, "beastd!"
[17:17] <iive> ramiro: he usually comes at night (central european time) and doesn't stay much after midnight.
[17:28] <burek> I'm stuck again
[17:28] <burek> I'm trying to build a static binary of ffmpeg
[17:29] <burek> so far I managed to successfully do this: http://pastebin.com/ewhuvR8G
[17:29] <burek> but as soon as i try to add this too: --extra-libs="-static -ldl" --extra-cflags=--static
[17:31] <ubitux> -static as extra lib?
[17:31] <ubitux> --static as cflags?
[17:31] <burek> well google told me that :D
[17:31] <ubitux> why are you trying to add -ldl?
[17:31] <ubitux> :/
[17:31] <ubitux> if that work, why are you trying to add flags?
[17:31] <burek> because I read in config.log that libdl is missing
[17:31] <burek> thats why
[17:31] <burek> it worked
[17:31] <burek> :D
[17:32] <burek> ok, what should i add
[17:32] <burek> just extra-ldflags?
[17:32] <ramiro> you shouldn't need extra-lib nor extra-cflags
[17:32] <burek> --extra-ldflags=-static
[17:32] <burek> or not even that?
[17:32] <ramiro> why do you need to add those flags?
[17:32] <burek> I'd like to build a static binary
[17:33] <burek> thats all
[17:33] <burek> which will be something like EXE file, so that i can copy it to another machine and just run it
[17:33] <burek> is that even possible on linux?
[17:34] <ramiro> hm, maybe you do need extra-ldflags. otherwise shared libraries will be picked up for the dependencies
[17:36] <ramiro> better yet would be to create a toolchain with just the static libraries you want as dependencies, that way there's no way a shared library will be picked up
[17:36] <burek> easy breezy  :D
[17:44] <burek> Ermine packs a GNU/Linux application together with any needed shared libraries and data files into a single executable.
[17:44] <burek> This file can be copied to any GNU/Linux host and run without further modifications.
[17:44] <burek> \o/ :)
[17:45] <burek> Free Trial... back to the lab..
[17:48] <ramiro> that looks like cheating...
[17:49] <burek> I don't give a crap :D
[17:49] <burek> I just want it to work :)
[19:51] <burek> ok, I've built static ffmpeg
[19:51] <burek> is there any volunteer to try it? :D
[19:53] <pasteeater> for linux?
[19:53] <burek> x86_64 yes :)
[19:54] <burek> http://pastebin.com/pRAVaFxX
[19:54] <burek> that's the configuration
[19:54] <burek> it has 30,232,576 bytes
[19:55] <burek> let me upload it if someone is crazy enough to actually test it :)
[19:59] <burek> here it is: http://ffmpeg.test-lab.ch/uploads/ffmpeg-static.tar.xz
[19:59] <burek> i just used statifier on the final ffmpeg binary and got that thing, which they claim is static
[20:10] <DaKaZ> all, I have a question about the implementation of the diff_y function in libavfilter/libmpcodecs/pullup.c
[20:11] <DaKaZ> why is there there an outer loop of 4&1 and what is the purpose of the a+=s; b+=s on line 221?
[20:34] <burek> this just doesn't work :S
[20:35] <burek> look guys, I'm ready to setup a machine to just build various static ffmpeg binaries, for people to have 1 binary to download and use
[20:35] <burek> without the need to build everything, everytime they need to do something with ffmpeg
[20:35] <burek> but I need help on how to correctly do this
[20:35] <durandal_1707> burek: what are you doing?
[20:36] <burek> I'm trying to make it easy for people to use ffmpeg, so that they can just download the binary on the machine (depending on their architecture) and run it
[20:36] <burek> cron job would rebuild binaries each day/week to make it always the latest
[20:37] <durandal_1707> burek: what does not work?
[20:37] <burek> so they don't need to compile it or to use package managers, which usually are outdated
[20:37] Action: durandal_1707 bored
[20:37] <pasteeater> burek: something like this? https://sites.google.com/site/linuxencoding/builds
[20:37] <burek> well I need directions on how to do it properly
[20:38] <burek> I love you man :D
[20:38] <burek> exactly :)
[20:38] <pasteeater> that's from relaxed
[20:38] <pasteeater> IIRC
[20:39] <burek> pasteeater, will that binary work on stable/testing release, along with sid?
[20:39] <burek> hack i;ll download it and try :D
[20:40] <pasteeater> i don't know
[20:40] <burek> thanks a lot for this :)
[20:40] <burek> that's exactly what I wanna do
[20:40] <burek> but more :)
[20:40] <burek> not just for sid
[20:41] <relaxed> burek: the requirements for my static builds are 64bit linux and a kernel >=2.6.26
[20:42] <burek> great job man :) it works
[20:42] <burek> is it possible to add all gpl libraries
[20:43] <burek> like libv4l2
[20:44] <durandal_1707> burek: it is possible, but dunno if it is redistributable
[20:45] <burek> that would be really great
[20:45] <burek> to include all gpl-ed libraries
[20:45] <burek> because a lot of people either don't know how to properly compile the ffmpeg
[20:45] <burek> or they use some outdated packages..
[20:46] <pasteeater> you won't want to use anything that requires --enable-nonfree for example (libfaac)
[20:46] <burek> no, not that
[20:46] <burek> just gpl
[20:46] <burek> i mean, i would make for myself libaacplus included :D
[20:46] <burek> but that's just for myself :)
[20:47] <burek> relaxed, does that binary work even on non-debian machines?
[20:47] <Daemon404> aac is one thing where the proprietary encoders are way better than all the FOSS ones
[20:47] <Daemon404> :V
[20:48] <burek> Daemon404, yes i agree :/
[20:48] Action: Daemon404 just uses qtaacenc
[20:48] <relaxed> burek: yes, as I've already stated it will work on any 64bit linux distro with a kernel >=2.6.26
[20:48] <burek> cool man :)
[20:49] <burek> I've been trying to do that for days :/
[21:03] <durandal_1707> michaelni: why our image regression tests create rgb24 samples instead of rgba like in libav?
[21:26] <michaelni> durandal_1707, id guess we probably fixed it first and libav wanted to do it differently, either way i dint think theres any reason for the difference
[21:27] <CIA-108> ffmpeg: 03Michael Niedermayer 07master * r247d30a7db 10ffmpeg/libavcodec/vp3.c: 
[21:27] <CIA-108> ffmpeg: vp3: Copy all 3 frames for thread updates.
[21:27] <CIA-108> ffmpeg: This fixes a double release of the current frame on deinit.
[21:27] <CIA-108> ffmpeg: Fixes CVE-2011-3934
[21:27] <CIA-108> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[21:27] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:27] <CIA-108> ffmpeg: 03Hendrik Leppkes 07master * r6071644287 10ffmpeg/libavcodec/indeo3.c: 
[21:27] <CIA-108> ffmpeg: indeo3: fix motion vector validation
[21:27] <CIA-108> ffmpeg: The index of the motion vector has to be checked before being
[21:27] <CIA-108> ffmpeg: multiplied by 2 for the array index.
[21:27] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:29] <ubitux> saste: i will need ffmpeg to be able to deal with audio filters soon
[21:29] <ubitux> saste: i'm basically able to generate a filterchain with the -map_channel options in ffmpeg.c
[21:29] <saste> ubitux: i'll send you my local patches, can't work on it before the weekend
[21:29] <ubitux> but of course, since ffmpeg has no capability to deal with an audio filtergraph
[21:29] <ubitux> it needs it first :p
[21:30] <ubitux> saste: ok :)
[21:30] <ubitux> saste: btw gitorious is really aweful
[21:30] <ubitux> i don't know how you are able to work with haha
[21:30] <saste> never tried an alternative, sticking with it was just cheaper, i'm lazy
[21:31] <ubitux> :)
[21:31] <Daemon404> github is decent
[21:31] <Daemon404> if you an get past the js bloat
[21:31] <ubitux> yes github is somehow good enough
[21:32] <Daemon404> its better than running your own git server
[21:32] <Daemon404> which is pretty painful 
[21:32] <Daemon404> (to set up)
[21:32] <ubitux> i had trouble with gitorious to just list the available branches
[21:32] <nevcairiel> setting up gitosis (or gitolite) is pretty easy imho
[21:32] <ubitux> it says "Branches: branch, foobar, xxxx, and 10 more", but you can't even clic "and 10 more"
[21:32] <Daemon404> nevcairiel, ever had to upgrade them?
[21:32] <Daemon404> ;)
[21:32] <ubitux> you need to go into one branch to list the other
[21:32] <nevcairiel> never touch a running system!
[21:33] <Daemon404> see
[21:33] <nevcairiel> my gitosis is ages old, still works
[21:33] <Daemon404> thats a shit mentality imho
[21:33] <ubitux> the painful part is not setting up gitosis/gitolite
[21:33] <nevcairiel> its just a management script, what would upgrading get me? <.<
[21:33] <ubitux> if you want to have public git it's painful
[21:33] <ubitux> you need to setup a web frontend
[21:33] <Daemon404> ^ this is what i mean
[21:33] <ubitux> and this means security constraints
[21:33] <nevcairiel> shrug, i didnt think it was that bad
[21:34] <nevcairiel> the web thing is read-only, what can go wrong
[21:34] <ubitux> haha :)
[21:34] <Daemon404> nevcairiel, people vary wildly in what they think is good/bad
[21:34] <Daemon404> some people even claim setting up a linux-based mail server isnt terribly painful
[21:34] <Daemon404> (tip: it is awful)
[21:34] <ubitux> nevcairiel: https://bugzilla.redhat.com/show_bug.cgi?id=479715
[21:34] <nevcairiel> that i can agree with, mail servers suck, especially if you want spam and virus checking
[21:35] <Daemon404> nevcairiel, ive also set up ms exchange servers
[21:35] <Daemon404> which were 100x easier
[21:35] <Daemon404> :V
[21:35] <Daemon404> than linux-based ones
[21:35] <nevcairiel> ubitux: well git itself is being updated regularly, of course!
[21:36] <ubitux> ?
[21:36] <ubitux> just replying to "the web thing is read-only, what can go wrong"
[21:36] <ubitux> everything can go wrong, especially with that kind of webapp
[21:36] <nevcairiel> "it could potentially have a security exploit" is a cheap reason for not running anything yourself. :d
[21:36] <Daemon404> no but it's a good reason for updating.
[21:36] <nevcairiel> git, yes
[21:36] <ubitux> it's the main reason i don't setup it for git
[21:37] <saste> ubitux: https://github.com/mnzaki/FFmpeg/commits/audio-filters-20110818
[21:37] <nevcairiel> gitosis is a server side script that just updates soem config files when i push to one specific repository
[21:37] <ubitux> and the main reason i'm still reluctant to do it for mails
[21:37] <Daemon404> whatabotu gitweb etc
[21:37] <ubitux> saste: oh rien, mnzaki branches, 'forgot about them
[21:37] <ubitux> thank you
[21:37] <ubitux> right*
[21:38] <nevcairiel> gitweb is part of git
[21:38] <Daemon404> it's third-party i believe.
[21:38] <Daemon404> got git core.
[21:38] <Daemon404> s/got/not/
[21:39] <nevcairiel> its part of the git distribution these days
[21:39] <ubitux> https://github.com/gitster/git/tree/master/gitweb
[21:39] <ubitux> it's in the main tree
[21:40] <nevcairiel> From the git version 1.4.0 gitweb is bundled with git. ^
[21:40] <ubitux> debian is certainly just splitting it like everything
[21:40] <Daemon404> ubitux, i dont mind that
[21:40] <Daemon404> i prefer it even
[22:27] <beastd> hi
[22:27] <ramiro> hi
[22:28] <beastd> Hey ramiro, how are you? Nice to see you around ^^
[22:29] <bcoudurier> hi ramiro 
[22:29] <iive> hi beastd  :P
[22:29] <ramiro> bcoudurier: hi
[22:30] <beastd> Oi  Ivan!
[00:00] --- Thu Jan 26 2012