[Ffmpeg-devel-irc] ffmpeg-devel.log.20130710

[burek]

[00:04] <saste> michaelni, what are dump_extra and remove_extra bsfs?
[00:04] <saste> is there some way to remove the global header from a file via a bsf?
[00:07] <michaelni> remove_extra removes the global header from the individual frames if it happens to be there
[00:07] <saste> michaelni, dump_extra, is any different?
[00:08] <saste> also how do we pass the args command by using the commandline?
[00:08] <saste> or was it never supported?
[00:09] <michaelni> dumo_extra adds the extradata in front of frames
[00:10] <saste> uhm ok
[00:10] <durandal_1707> saste: so i should use av_strtok only?
[00:10] <saste> as for what regards the argument, do you confirm it can't be passed to the ff* tools right now?
[00:10] <saste> durandal_1707, you should be able to use av_strtok only
[00:11] <saste> but note that Nicholas added support to lists in opt.h
[00:11] <saste> but i'm not sure they're good enough for you
[00:14] <durandal_1707> i don't see them
[00:16] <michaelni> saste, i dont see how it could be passed either
[00:17] <saste> durandal: av_opt_set_int_list
[00:23] <durandal_1707> i fail to see how i could use this
[00:36] <cone-395> ffmpeg.git 03Michael Niedermayer 07master:3aa2257d240a: doc/RELEASE_NOTES: add a note about AVISynth
[00:36] <cone-395> ffmpeg.git 03Alexis Ballier 07master:c3ca70204bd8: codec_desc: fix typo in rv20 long_name.
[00:39] <johnsmith_> Has anyone here successfully compiled the latest FFMPEG version on centos 5?
[00:40] <durandal_1707> there is problem?
[00:41] <johnsmith_> Been spending at least 10 hours trying to get the latest FFMPEG onto Centos 5
[00:41] <johnsmith_> the yum install ffmpeg ffmpeg-devel has deprecated functions only
[00:42] <Compn> what is the problem
[00:42] <Compn> build error? 
[00:42] <Compn> paste it to pastebin
[00:43] <Compn> because a) none of us are using centos5 and b) we cant read minds, especially computer minds
[00:43] <johnsmith_> Hmm..
[00:44] <johnsmith_> One of the symptoms I have is libavcodec.so.52 is actually missing within usr/local/libs
[00:45] <johnsmith_> And another is problems getting references to av_parser from the util.c files
[00:46] <johnsmith_> I think the it's really hard to pinpoint the exact problem of the build
[00:47] <johnsmith_> I am following the centos compilation guide, tweaking the fdk-aac and omitting opus libs
[00:47] <johnsmith_> then running ffmpeg
[00:48] <johnsmith_> ffmpeg successfully does a make install operation
[00:48] <johnsmith_> but gcc/eclipse yell at me with those problems mentioned above
[00:48] <durandal_1707> libavcodec.so.52 is very old, 55 is latest one
[00:50] <johnsmith_> Should it even exist within usr/local/libs in the first place?
[00:52] <durandal_1707> if you have old one installed...
[00:53] <johnsmith__> Here is the eclipse output
[00:53] <johnsmith__> http://pastebin.com/5cVeNCea
[00:53] <durandal_1707> this is not #eclipse
[00:54] <johnsmith__> this is the output trying to compile it
[00:54] <durandal_1707> why you use that to compile?
[00:54] <durandal_1707> that is not supported way to compile ffmpeg
[00:54] <johnsmith__> i can use gcc to do it as well
[00:54] <johnsmith__> give me a sec
[00:55] <johnsmith__> http://pastebin.com/9mCNHP39
[00:55] <johnsmith__> same thing
[01:04] <cone-395> ffmpeg.git 03Marton Balint 07master:3886572a83bf: ffplay: do not update audio clock with old audio pts if current frame is AV_NOPTS_VALUE
[01:04] <cone-395> ffmpeg.git 03Marton Balint 07master:f07cb53ab951: ffplay: assume 0 stream start time if start time is unset in duration check
[01:04] <cone-395> ffmpeg.git 03Michael Niedermayer 07master:49f441ea8991: MAINTAINERS: Update which releases are supported
[01:04] <cone-395> ffmpeg.git 03Michael Niedermayer 07master:9e214900ad31: Merge remote-tracking branch 'cus/stable'
[01:15] <durandal11707> what why 2.0?
[01:15] <durandal11707> what is wrong with 1.4 ?
[01:18] <durandal11707> perhaps because of refcounting?
[01:20] <cone-395> ffmpeg.git 03Eugene Dzhurinsky 07master:09f59d6adfa1: avformat/gif: Do not write GIF89a header if loop is set to -1.
[02:06] <johnsmith> durandal, thanks for giving me the hint that the libavcodec.so.25 was an old file, using old methods required old files
[02:09] <johnsmith> I can successfully compile the ffmpeg decoding encoding example on the website, but I realized that me calling avformat_find_stream_info(..) caused all sorts of reference issues
[02:09] <johnsmith> is it because that method is deprecated?
[02:38] <cone-395> ffmpeg.git 03Luca Barbato 07master:5eaed6d33689: atrac3: fix error handling
[02:38] <cone-395> ffmpeg.git 03Michael Niedermayer 07master:1ebb95d5f193: Update for 2.0
[02:41] <cone-395> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/heads/release/2.0': unknown revision or path not in the working tree.
[02:41] <cone-395> Use '--' to separate paths from revisions
[02:41] <cone-395> refs/heads/release/2.0:HEAD: Update for 2.0
[02:49] <cone-395> ffmpeg.git 03Michael Niedermayer 07master:2b8b2ba19fe0: Changelog: replace next by 2.0
[02:49] <cone-395> ffmpeg.git 03Michael Niedermayer 07release/2.0:2b8b2ba19fe0: Changelog: replace next by 2.0
[03:12] <cone-395> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/tags/n2.0': unknown revision or path not in the working tree.
[03:12] <cone-395> Use '--' to separate paths from revisions
[03:12] <cone-395> refs/tags/n2.0:HEAD: Changelog: replace next by 2.0
[03:43] <cone-395> ffmpeg.git 03Michael Niedermayer 07master:a37e42b3ee42: Update back for git master
[03:43] <cone-395> ffmpeg.git 03Michael Niedermayer 07master:0574fe76b857: Changelog: add back the <next> placeholder for new changes
[03:43] <cone-395> ffmpeg.git 03Piotr Bandurski 07master:ccf9211e29bd: avformat/utils: avformat_find_stream_info fix a crash in case of oom
[08:42] <mark4o> Shouldn't the Changelog link on the download.html page link to the Changelog ( http://git.videolan.org/?p=ffmpeg.git;a=blob_plain;f=Changelog;h=n2.0 )?
[08:42] <mark4o> or maybe two links, Changelog and git shortlog
[08:43] <mark4o> oops I mean http://git.videolan.org/?p=ffmpeg.git;a=blob_plain;f=Changelog;hb=n2.0
[10:23] <Bor0> what is the best way to get current frame? currentpts*fps works but not too good
[10:31] <av500> ?
[10:50] <xlinkz0> saste: i am ready to receive my robe and wizard hat
[10:58] <xlinkz0> can i disable HAVE_TERMIOS_H ?
[11:00] <saste> xlinkz0, what?
[11:00] <xlinkz0> you said yesterday that i can't possibly make ffmpeg into a library.. unless i was a wizard
[11:01] <xlinkz0> and so it is done :D
[11:01] <saste> xlinkz0, what did you do exactly?
[11:02] <xlinkz0> saste: a library that facilitates this : http://codepad.org/wFaMWG3K
[11:03] <xlinkz0> now i need to test it on ios/android to make sure it works but i don't see why it wouldn't
[11:03] <xlinkz0> since the libav libs work..
[11:03] <saste> xlinkz0, so in android/ios you can't use system()?
[11:03] <xlinkz0> no
[11:04] <xlinkz0> you can not create processes unless you root the phone
[11:04] <saste> uh i see
[11:06] <xlinkz0> only one else that done this charges 400$ per licence lol
[11:06] <xlinkz0> http://ffmpeg4android.netcompss.com/home/purchase
[11:07] <xlinkz0> are they allowed to sell it?
[11:08] <atiti> lol
[11:08] <JEEB> as long as they comply with the software license
[11:08] <atiti> theyr probably selling the app that runs the command
[11:08] <xlinkz0> you can not launch other apps from inside your app
[11:08] <atiti> which is basically a terminal app
[11:09] <xlinkz0> no such thing as far as i understand
[11:10] <atiti> xlinkz0: its not launching another app, its executing the ffmpeg binary 
[11:10] <atiti> at least as far as i can see on the screenshot
[11:10] <xlinkz0> which is not possible from normal apps unless the user roots the phone
[11:10] <atiti> http://stackoverflow.com/questions/2843250/how-to-run-terminal-command-in-android-application
[11:10] <xlinkz0> either this is false or my phone dev colleagues are n00bs :(
[11:11] <xlinkz0> even if it works for android it won't work on ios
[11:11] <atiti> sure its possible without rooting
[11:11] <JEEB> the app just has to be in a place where it can have the execution bit, IIRC?
[11:11] <atiti> all apps run as a user, so you can easily execute commands under that user
[11:11] <xlinkz0> afaik you need to ship the ffmpeg binary with the app
[11:11] <atiti> rooting is what the name stands for, how to get root on the phone :p
[11:12] <xlinkz0> and when you ship it you must  be able to chmod it which you can not..
[11:12] <atiti> yes, you can chmod it if your own user owns it
[11:12] <xlinkz0> so you may be able to run ls which is already chmoded
[11:12] <atiti> which it does, if it comes with your app
[11:13] <xlinkz0> you tested it?
[11:13] <JEEB> anyways, not sure how they do it but it sure looks like launching the command line interface in one way or another. Granted, I don't take out the possibility of herping a derp and actually making the CLI a "lib" :D
[11:13] <atiti> xlinkz0: yes
[11:13] <xlinkz0> well then i'll just go kill the android dev
[11:14] <xlinkz0> hope this doesn't work on ios atleast, the ios dev said he can't call other processes
[11:15] <atiti> im not familiar with the ios buildup so i cant comment on that
[11:17] <xlinkz0> <Kanalia> xlinkz0: you can't run external executables/binaries from your app
[11:17] <xlinkz0> mwhahaha
[11:17] <xlinkz0> my work was not in vain
[11:17] <xlinkz0> (ios)
[12:24] <cone-59> ffmpeg.git 03Luca Barbato 07master:874c8a17ac9b: atrac3: fix error handling
[12:24] <cone-59> ffmpeg.git 03Luca Barbato 07master:22e76ec635ba: atrac3: set the getbits context the right buffer_end
[12:24] <cone-59> ffmpeg.git 03Michael Niedermayer 07master:40e8967ee161: Merge commit '22e76ec635bafdd1d1ec35581a7ac09e69e3c43e'
[12:36] <cone-59> ffmpeg.git 03Luca Barbato 07master:50cf5a7fb788: atrac3: Error on impossible encoding/channel combinations
[12:36] <cone-59> ffmpeg.git 03Luca Barbato 07master:bbf6a4aa20bf: imc: Catch a division by zero
[12:36] <cone-59> ffmpeg.git 03Michael Niedermayer 07master:a3d9a2168355: Merge commit 'bbf6a4aa20bfe3d7869b2218e66063602dfb8aa7'
[13:08] <cone-59> ffmpeg.git 03Luca Barbato 07master:12576afe206d: adpcm: Write the correct number of samples for ima-dk4
[13:08] <cone-59> ffmpeg.git 03Michael Niedermayer 07master:067c9c3b6319: Merge commit '12576afe206d35231ccd61f9033c5fdab6a11e80'
[13:48] <durandal_1707> michaelni: why 2.0 is not put into news?
[13:49] <durandal_1707> and why "Nameless" as name, there are much better names....
[14:09] <cone-59> ffmpeg.git 03Paul B Mahol 07master:410192e36958: add crc24 tables
[14:09] <cone-59> ffmpeg.git 03Paul B Mahol 07master:8491f4050053: tak: use crc table from lavu
[14:09] <cone-59> ffmpeg.git 03Paul B Mahol 07master:4977e467a50a: lavf/takdec: check metadata blocks checksum
[14:28] <kierank> ubitux: your r128 filter was presented today
[14:38] <cone-59> ffmpeg.git 03Paul B Mahol 07master:884c890535c5: lavfi: add aecho filter
[14:44] <durandal_1707> kierank: on cnn?
[14:44] <kierank> at rmll
[14:46] <durandal_1707> video?
[14:47] <Compn> was it for ffmpeg talk or used in a different project ? 
[15:06] <cone-59> ffmpeg.git 03Michael Niedermayer 07master:2c0ec526e2c3: fate/crc fix after master:410192e36958: add crc24 tables
[15:14] <cone-59> ffmpeg.git 03Luca Barbato 07master:adadc3f2443d: qdm2: refactor joined stereo support
[15:14] <cone-59> ffmpeg.git 03Michael Niedermayer 07master:564909093a31: Merge commit 'adadc3f2443d25b375e21e801516ccfd78e0b080'
[15:20] <cone-59> ffmpeg.git 03Paul B Mahol 07master:bc95b9428950: lavfi/aconvert: unbreak
[15:37] <durandal_1707> michaelni: earwax filter overreads & overwrites
[15:40] <durandal_1707> asetnsamples=17,earwax
[16:07] <cone-59> ffmpeg.git 03Luca Barbato 07master:4ecdb5ed4459: qdm2: Conceal broken samples
[16:07] <cone-59> ffmpeg.git 03Michael Niedermayer 07master:2e6338b47276: Merge remote-tracking branch 'qatar/master'
[16:30] <ghooman> Hi, i'm trying to make a custom audio filter
[16:33] <ghooman> can i retrieve file duration from filter context&
[16:35] <durandal_1707> ghooman: nope
[16:35] <durandal_1707> why would you need it?
[16:38] <ghooman> to calculate average volume on 1/1800
[16:38] <durandal_1707> 1/1800 ?
[16:38] <ghooman> part of
[16:39] <durandal_1707> you can count number of samples
[16:41] <ghooman> i want to produce 1800(or some other number) of json values for http://www.waveformjs.org/
[16:43] <ghooman> i need that value at start, because i want to calculate size of 1/n segment
[16:45] <durandal_1707> there is no way to get duration from filter context
[16:45] <ghooman> thx
[16:46] <durandal_1707> but as 1800 is fixed, when it does not need to be, which means you can get duration via other means and pass it to filter
[16:47] <cone-59> ffmpeg.git 03Piotr Bandurski 07master:b0509563347b: avformat/utils: avformat_find_stream_info set value for ret in case of oom
[16:48] <cone-59> ffmpeg.git 03Michael Niedermayer 07master:fbe159e85079: avcodec/qdm2: store bits in an integer instead of float variable
[16:48] <cone-59> ffmpeg.git 03Michael Niedermayer 07master:8f09957194b8: avcodec/qdm2: initialize sign_bits
[16:48] <cone-59> ffmpeg.git 03Michael Niedermayer 07master:0a3a0edd52b9: avfilter/af_earwax: Fix out of array accesses on odd packets
[17:03] <mateo`_> hi guys, anyone to review/comment my lavf/movenc patch ?
[17:19] <ubitux> kierank: fun, any feedback?
[17:20] <michaelni> ubitux, do you want to review mateo`_ patch  or that webvtt patch ?
[17:20] <ubitux> i'm sorry i can't currently
[17:20] <ubitux> i'll be available full time in august for ffmpeg
[17:20] <michaelni> ok
[17:20] <ubitux> but nothing until then
[17:21] <saste> michaelni, what about the names of psychedelic drugs?
[17:23] <michaelni> saste, i dont think thats a good idea
[17:24] <saste> michaelni, it doesn't have to be a good idea, must be fun
[17:26] <ubitux> http://namingschemes.com/
[17:31] <durandal_1707> michaelni: for next releases, just do nameless II, nameless III and so on
[17:31] <saste> nameless unreleased
[17:32] <durandal_1707> http://namingschemes.com/Vagina_Synonyms
[17:33] <ubitux> that will certainly bring a lot of female contributors
[17:38] <durandal_1707> only this remain: http://namingschemes.com/Pokemon
[17:46] <aballier> michaelni: have you considered dropping 1.1 branch and still keeping 1.0 maintained ? (I dont see any reason to use 1.1 but not 1.2 however 1.0 is nice for downstreams since it doesnt have the planar formats)
[18:08] <michaelni> aballier, would be a possibility if thats what people prefer, but maintaining older releases is more work than more recent ones. 
[18:09] <michaelni> If someone wants to take over maintaince of a old release or wants to help that is welcome
[18:14] <cptspiff> regarding topic. will it alternative libav/conflict libav or is the lib name pissing contest settled? former would require abi compat which sounds a bit.. challenging. or do you just want the fluff for third party repos?
[18:15] <Compn> cptspiff : which topic ? which naming conflicts ?
[18:15] <cptspiff> debian packaging
[18:16] <Compn> you mean libav masquerading as 'ffmpeg' package ? its up to debian
[18:16] <Compn> or ffmpeg trademark owner...
[18:17] <j-b> FLV1 and MPEG-4 ASP have something in common?
[18:17] <cptspiff> debian wont accept several packages  with conflicting filenames unless they are alternatives
[18:17] <Compn> j-b : i thought h263 , have to check what flv1 codec relies on
[18:18] <Compn> cptspiff : well i think saste was trying to discuss with ubuntu/debian. i dont know outcome of that
[18:18] <cptspiff> k. anyways just waned to offer my fu..
[18:19] <Compn> j-b : yes , flv links against mpegvideo and h263, same as mpeg4 asp ...
[18:19] <michaelni> cptspiff, both variants, as alternative as well as with different soname could be made, its not that one of the 2 needs to be picked ad the other could then not be done
[18:20] <j-b> Compn: thx
[18:20] <michaelni> cptspiff, about debian/ubuntu also talk with beastd 
[18:26] <aballier> michaelni: I could try to help but i'm not sure if i'm competent enough; I was mainly asking because maintaining 1.1 seemed not worth the work
[18:28] <aballier> michaelni: on a semi-related topic, are the recent jpeg2000 fixes needed for the old j2k code in 1.0 ?
[18:31] <michaelni> hard to say, some possibly might be needed, but j2kdec was marked as experiemntal so it should not have been used by default but then some applications might have enabled experimental decoders by default on their own
[18:32] <michaelni> about "compent" if you maintain 1.0 or help it will be in better shape than if you dont
[18:33] <j-b> http://xhevc.com/en/downloads/downloadCenter.jsp based on ffmpeg 1.0.6
[18:33] <j-b> and there is source in it
[18:37] <aballier> one possibility could be to disable j2kdec at build time and rely on openjpeg
[18:37] <michaelni> yes
[18:37] <aballier> michaelni: when did you stop backporting fixes to 1.0 ?
[18:37] <aballier> (so that i know where to start from in e.g. 1.2 branch)
[18:44] <michaelni> the last release from 1.0 was 1.0.7 from 2013-05-14, after that i thnk there are just some bugfixes backported by carl
[18:45] <aballier> thanks, will have a look
[19:04] <kierank> ubitux: true peak and hardcoded 48khz would be nice according to the guy
[19:04] <ubitux> hardcoded 48khz ?
[19:05] <ubitux> you mean removing the hardcoded 48khz to support other freq?
[19:05] <kierank> yes
[20:24] <cone-59> ffmpeg.git 03Carl Eugen Hoyos 07master:c9eb5c9751c8: lavf/utils.c: Avoid a null pointer dereference on oom after duration_error allocation.
[20:24] <cone-59> ffmpeg.git 03Carl Eugen Hoyos 07master:1db88c33f2c0: Suggest recompilation with openssl or gnutls if the https protocol is not found.
[20:24] <cone-59> ffmpeg.git 03Michael Niedermayer 07master:07206deed103: Merge remote-tracking branch 'cehoyos/master'
[20:46] <mark4o> durandal_1707: [-]HH[:MM[:SS[.m...]]] ?   should be [-][HH:]MM:SS[.m...], yes?
[20:46] <mark4o> looks like other parts of doc have it wrong also
[20:48] <ubitux> i would say [-][[HH:]MM:]SS[.m...]
[20:48] <durandal_1707> patch welcome
[20:48] <mark4o> well it also has [-]S+[.m...]
[20:49] <mark4o> durandal_1707: a patch to your patch?
[20:50] <durandal_1707> i can fix it in my patch
[21:10] <durandal_1707> mark4o: i mean i will fix hhmmss in code i add
[21:11] <cone-59> ffmpeg.git 03Paul B Mahol 07master:45119739019e: wmalosslessdec: do not set avctx->coded_frame
[21:12] <cone-59> ffmpeg.git 03Derek Buitenhuis 07master:2ea60971dca9: fate: Always enable avresample
[21:14] <cone-59> ffmpeg.git 03Paul B Mahol 07master:1e89f74902c6: lavfi/afade: use av_rescale()
[21:24] <Daemon404> michaelni, when you say "more complex things than read a variable" for teh avio callback, what do you mean?
[21:24] <Daemon404> i do something a bit more complex, but it may or may not be classified as "read a variable"
[21:25] <Daemon404> https://github.com/dwbuiten/d2vsource/blob/master/core/decode.cpp#L44 and the following func, for referene
[21:27] <cone-59> ffmpeg.git 03Paul B Mahol 07master:badbd2414d38: lavfi/aevalsrc: use AV_OPT_TYPE_DURATION
[21:27] <cone-59> ffmpeg.git 03Paul B Mahol 07master:db4e191d8897: lavfi/aevalsrc: do not free AVOption variables in uninit()
[21:59] <michaelni> Daemon404, i dont see any usage of the interrupt callback in there
[22:00] <Daemon404> michaelni, uau already replied on the mL
[22:02] <durandal_1707> that's not uau iirc
[22:02] <nevcairiel> no its not
[22:02] <Daemon404> woops
[22:02] <Daemon404> youre  right
[22:02] <durandal_1707> s/that's/he's ...
[23:23] <j-b> michaelni: is there a way to disable the VirtualAlloc/Free calls from swscale ? A config option?
[23:26] <cehoyos> j-b: Did you try --disable-mmxext ?
[23:28] <j-b> cehoyos: o_O
[23:29] <cehoyos> Am I wrong?
[23:31] <j-b> cehoyos: probably not
[23:31] <cehoyos> (I just don't understand the emoticon and Google doesn't really help.)
[23:34] <michaelni> j-b, i can add a configure option probably to disable it
[23:35] <j-b> no, I just was wondering about it
[23:35] <j-b> I'll find another way
[23:38] <durandal_1707> is there reasonable way to automaticaly free pads that get dynamically allocated?
[23:40] <durandal_1707> i could add some flag to each pad that get allocated....
[00:00] --- Thu Jul 11 2013