[Ffmpeg-devel-irc] ffmpeg-devel.log.20121116

[burek]

[01:27] <Compn> unicode
[01:28] <Compn> try to copy and paste a unicode script from web browser to email :P
[01:28] <Compn> and then to terminal
[05:21] <cone-195> ffmpeg.git 03Michael Niedermayer 077845f8d282a9: vc1dec: do not allow field_mode to change after the first header
[05:21] <cone-195> ffmpeg.git 03Michael Niedermayer 0768def00a6330: rv34: check image size before using it
[09:32] <Tjoppen> wups, missed the updated mxfdec patch
[09:39] <ubitux> j-b: it's curious you don't mention the "company copyright"
[09:39] <ubitux> in your post(s)
[09:39] <ubitux> unless i'm mistaken
[09:39] <j-b> company copyright?
[09:39] <ubitux> for instance someone who write some code for a company, the code doesn't belong to him, but to the company (under french law at least)
[09:40] <ubitux> even if there is his name
[09:40] <ubitux> so he/you have to contact the company itself
[09:40] <j-b> Yes, because I did not find it important enough to speak of it
[09:40] <j-b> We had 5 times the case
[09:40] <ubitux> ok :)
[09:41] <ubitux> i thought it might have been more complicated than contacting the author himself
[09:41] <j-b> ubitux: No
[09:41] <Daemon404> id think it'd be easier
[09:41] <ubitux> and more "dangerous"
[09:41] <Daemon404> companies hate gp
[09:41] <Daemon404> l
[09:42] <j-b> very few people used their corporate companies email
[09:42] <j-b> but, BBC, SFR, Jutel are the companies I am quickly thinking about
[09:42] <ubitux> Daemon404: or don't care, and could want to sue you because you relicensed things to get money :p
[09:42] <ubitux> ok
[09:42] <j-b> ubitux: I will speak about that in the 3rd part
[09:43] <ubitux> your posts are missing graphs!
[09:43] <j-b> ubitux: and you will see why I do not care if I am not 100% sure of my relicensing
[09:43] <ubitux> http://mathematica.stackexchange.com/questions/11350/xkcd-style-graphs  don't forget to add this in part3
[09:43] <j-b> ubitux: my posts are done quickly
[09:43] <ubitux> :)
[09:43] <j-b> graph about what?
[09:43] <ubitux> how much replied, how much ok'ed etc
[09:43] <j-b> percentage of assholes in past VLC contributors?
[09:43] <Daemon404> ubitux, those dont look right
[09:43] <Daemon404> not like xkcd
[09:43] <Daemon404> which iirc are hand drawn
[09:44] <j-b> all replied but a few
[09:44] <j-b> graph would be boring
[09:44] <Daemon404> percentage of death threats whiel relicensing
[09:44] <ubitux> i'm just joking, don't take this too seriously :p
[09:44] Action: Daemon404 runs
[09:54] <ubitux> j-b: what happens to GPL parts too huge/important where relicensing is refused?
[09:54] <ubitux> do they stay in, with conditional compilation?
[09:56] <j-b> yes
[09:56] <j-b> and they will be replaced one by one
[09:57] <av500> hacker news is where GPL goes to die
[09:57] <ubitux> j-b: is that really a problem to have some parts in gpl?
[09:58] <j-b> ubitux: depends.
[09:58] <ubitux> (if disabled they don't hurt that much the project)
[09:58] <j-b> but one has to maintain them
[09:58] <ubitux> what do you mean?
[09:58] <j-b> ubitux: but so far, there is MPC demuxer, TiVO demuxer that are annoying. And mono filter.
[09:59] <ubitux> yadif? :p
[09:59] <j-b> The rest is useless or not critical, like yadif and hqnd3d
[09:59] <j-b> yadif is non-critical
[09:59] <ubitux> what's mono filter?
[10:00] <j-b> downmixing to mono with headphone enhancments
[10:00] <j-b> or somehting
[10:00] <ubitux> ok
[10:00] <j-b> for mono filter, well, libavresample and libswresample will do the job
[10:00] <j-b> for MPC and TiVO, I will go on the effort
[10:01] <ubitux> non-response, or denied?
[10:01] <j-b> non-response
[10:01] <j-b> Tivo is harder, because it comes from the mplayer-tivo project
[10:02] <Daemon404> the mono thing can be replaced by one of the resampling libs here cant it?
[10:02] <Daemon404> which are lgpl
[10:02] <ubitux> (isn't mpc supported by ffmpeg?)
[10:02] <Daemon404> it is, but inb4 busted
[10:02] <j-b> 10:00 <@j-b> for mono filter, well, libavresample and libswresample will do the job
[10:02] <ubitux> why not just improve this one?
[10:03] <j-b> that is basically what is going to happen
[10:03] <ubitux> j-b contributing to lavc \o/
[10:03] <Daemon404> lavf*
[10:03] <j-b> ubitux: so, probably a --disable-gpl flag will appear
[10:03] <ubitux> :)
[10:04] <Daemon404> then you can argue over what is gpl compatible in configure
[10:04] <Daemon404> just like us
[10:04] <Daemon404> :)
[10:04] <j-b> and dead-demuxers will die
[10:04] <ubitux> and then you realize the configure is in GPL
[10:04] <Daemon404> im still waiting for liba52 to go away
[10:04] <Daemon404> alogn with libmad
[10:04] <Daemon404> i cant remember why vlc uses  those
[10:05] <j-b> liba52 because it is used as an audio-filter, hence works with SPDIF
[10:05] <j-b> libmad because its quality is better.
[10:05] <Daemon404> couldnt that be replaced by the spdif muxer in lavf?
[10:05] <Daemon404> or something
[10:06] <ubitux> (related: http://www-archive.mozilla.org/MPL/relicensing-faq.html http://blog.gerv.net/2006/03/relicensing_complete/)
[10:06] <j-b> Daemon404: no.
[10:06] <Daemon404> any reason?
[10:06] <Daemon404> (yes, in uninformed)
[10:06] <Daemon404> s/in/im/
[10:06] <j-b> 10:05 <@j-b> liba52 because it is used as an audio-filter, hence works with SPDIF
[10:07] <j-b> we are looking to drop that
[10:07] <j-b> but so far, not possible.
[10:07] <Daemon404> im not seeing what it can do that lavf's spdifmuxer couldnt?
[10:07] <Daemon404> or is this for interfacing directly
[10:07] <j-b> yes. this is for live switching
[10:07] <Daemon404> (iirc, nevcairiel uses it to bitsteam)
[10:07] <j-b> and the issue is not in lavc, nor lavf, but in VLC architecture
[10:08] <Daemon404> oh
[10:09] <j-b> SPDIF and HDMI need to be audio filters, to change them live
[10:09] <Daemon404> in theory, once could write a minimal spdif filter using lavf
[10:10] <Daemon404> in practice, the libav* interlibrary deps are a bithc
[10:10] <j-b> Anyway, ubitux, the playback stuff are all lgpl, as you can see.
[10:10] <ubitux> ok ok
[10:11] <j-b> If they are not GPL, then, there is a FFmpeg fallback
[10:11] <j-b> except mono, for now.
[10:12] <Daemon404> mono is extremely important symbian playback.
[10:12] Action: Daemon404 runs
[10:12] <j-b> Daemon404: for iOS playback.
[10:13] <Daemon404> really?
[10:13] <j-b> Yes.
[10:13] <Daemon404> hah.
[10:13] <ubitux> does it do that much more advanced things than a simple downmix?
[10:14] <j-b> Mono, no. But the headphones yes.
[10:15] <j-b> http://git.videolan.org/?p=vlc.git;a=blob;f=modules/audio_filter/channel_mixer/headphone.c;hb=HEAD
[10:15] <j-b> http://git.videolan.org/?p=vlc.git;a=blob;f=modules/audio_filter/channel_mixer/dolby.c;hb=HEAD
[10:15] <Daemon404> that is a bizarre coding style
[10:16] <j-b> and yet, I am sure you can cope with it :)
[10:16] <ubitux> swresample.c:    { "dolby", "select Dolby",              0, AV_OPT_TYPE_CONST, { .i64 = AV_MATRIX_ENCODING_DOLBY }, INT_MIN, INT_MAX, PARAM, "matrix_encoding" },
[10:17] <ubitux> maybe you can add a "headphone" profile to honor in the rematrix, just like this one
[10:17] <Daemon404> ... that is extremely non-descriptive
[10:17] <j-b> ubitux: hence I said it was doable to do something
[10:18] <ubitux> clearly
[10:18] <j-b> And nothing was critical
[10:18] <ubitux> Daemon404: select <matrix_encoding> Dolby mode sounds ok to me
[10:19] <j-b> ubitux: however, you are right yadif would be nice.
[10:19] <j-b> but not mandatory for an video player
[10:19] Action: Daemon404 feels so 2003 when yadif is mentioned
[10:19] <j-b> Daemon404: do you have something better?
[10:20] <Daemon404> yes, but obviously not real-time
[10:20] <Daemon404> oh well
[10:20] <Daemon404> just yes
[10:20] <j-b> :)
[10:20] <j-b> liar
[10:20] <ubitux> j-b: did you see the guy on ffmpeg-devel recently talking about funding the relicencing?
[10:20] <Daemon404> i have baptistes port of handbrake's decomb filter
[10:20] <Daemon404> to lavfi
[10:20] <j-b> decomb is better?
[10:21] <Daemon404> it mixes a buncxh of things
[10:21] <Daemon404> it is supposedly better
[10:21] Action: Daemon404 poitns at j45 
[10:21] <j-b> license is?
[10:21] <Daemon404> whatever it is in handbrake... 
[10:21] <Daemon404> so i dunno ;p
[10:21] <Daemon404> i have to fish it out of my email
[10:22] <Daemon404> there's also nnedi3... but i think tahts gpl only
[10:22] <Daemon404> and slow.
[10:22] <ubitux> j-b: http://lists.ffmpeg.org/pipermail/ffmpeg-devel/2012-October/133288.html
[10:23] <ubitux> j-b: and the following thread: http://lists.ffmpeg.org/pipermail/ffmpeg-devel/2012-November/133448.html
[10:23] <j-b> well, I'll look at it later :)
[10:50] <cone-480> ffmpeg.git 03Nicolas George 07468781f8287b: doc/APIchanges: clarify comment on new APIs.
[10:50] <cone-480> ffmpeg.git 03Nicolas George 07926481078b4f: lavu/opt: fix av_opt_get_key_value() API.
[11:32] <ubitux> it's fun how LQ snow encodes actually look like snow :D
[12:31] <pettter> is there any nice and easy rundown of what timestamps to set, to what, and at what point, when encoding and muxing both audio and video into a container, where all three have different time_bases?
[13:16] <cone-480> ffmpeg.git 03Diego Biurrun 07e28b7553aeab: configure: Add option to disable all command line programs
[13:16] <cone-480> ffmpeg.git 03Janne Grunau 07116ae7285e2b: doxygen: remove obsolete options from Doxyfile
[13:16] <cone-480> ffmpeg.git 03Martin Storsjö 079eded0fe412e: configure: Check for -Werror parameters on clang
[13:16] <cone-480> ffmpeg.git 03Mans Rullgard 07ada51a334aeb: avserver: remove daemon mode
[13:16] <cone-480> ffmpeg.git 03Janne Grunau 07a4cd2ad89aa6: avserver: use freopen to redirect stdin/out/err to /dev/null
[13:16] <cone-480> ffmpeg.git 03Janne Grunau 07c84cce5a99b8: mxfdec: fix typo in mxf_read_seek()
[13:16] <cone-480> ffmpeg.git 03Janne Grunau 07bf5f46b4cc47: APIChanges: add entry for av_read_packet deprecation
[13:16] <cone-480> ffmpeg.git 03Michael Niedermayer 073db32ec119e5: Merge commit 'bf5f46b4cc47b7a4568119f224057d4ff91b6cdd'
[13:16] <cone-480> ffmpeg.git 03Michael Niedermayer 07f30cf51d4634: ffserver: put NoDaemon back so as not not break existing conf files.
[13:21] <Compn> did libav stop logging libav-devel ? i dont see a ml for it anymore
[13:22] <Compn> i mean the libav-devel-irc
[13:23] <ubitux> no idea
[13:23] <ubitux> channel is public, you can join
[13:29] <Compn> well you say that
[13:29] Action: Compn wonders if michaelni can join
[13:33] <ubitux> 13:33:00 ::: 1 - #libav-devel: ban *!*~michael*@*.4.15.vie.surfer.at [by wolfe.freenode.net, 16473203 secs ago]
[13:33] <ubitux> i don't think so
[13:34] <durandal_1707> ubitux: hmm, you have that old logs? or you saved just that line ?
[13:35] <ubitux> it's the current banlist
[13:35] <ubitux> i'm on the channel
[13:36] <cone-480> ffmpeg.git 03Janne Grunau 0739c4afd92694: avserver: fix constness casting warnings
[13:36] <cone-480> ffmpeg.git 03Janne Grunau 07566de8cde29d: avserver: replace av_read_packet with av_read_frame
[13:36] <cone-480> ffmpeg.git 03Diego Biurrun 078ff0f7763d98: crc: av_crc() parameter names should match between .c, .h and doxygen
[13:36] <cone-480> ffmpeg.git 03Diego Biurrun 078c3849bc76c1: x86: dsputil: port to cpuflags
[13:36] <cone-480> ffmpeg.git 03Michael Niedermayer 07ff3b59c848b6: Merge remote-tracking branch 'qatar/master'
[13:37] <durandal_1707> omg, i got spam with stupid irc command
[13:38] <Compn> ubitux : i was just curious how michael is merging changes from libav without seeing half of the discussion. makes it more difficult 
[13:39] <Compn> both ffmpeg and libav patches say 'as discussed on irc' or some such 
[13:39] <ubitux> they don't discuss that much technical stuff on irc
[13:39] <ubitux> Compn: they are not discussed
[13:39] <Compn> just the flames then ?
[13:39] <Compn> eh
[13:40] <ubitux> generally, there is just one developer getting pissed of waiting for review and hassle another for a ok
[13:40] <Compn> alright
[13:40] <Compn> good to know
[13:40] <ubitux> but yeah sometimes they say some stuff important
[13:40] <durandal_1707> it is heaven for trolls
[14:11] <cone-480> ffmpeg.git 03Clément BSsch 072dd0da787ce5: lavf/mp3enc: make sure the header is valid before writing audio pkt.
[14:16] <cone-480> ffmpeg.git 03Xidorn Quan 078ee6db393c92: mov: add more udta meta data recognition
[14:16] <cone-480> ffmpeg.git 03Matthieu Bouron 0783cab07a4c7f: mxfdec: set audio packets pts
[16:36] <cbsrobot> j-b: have you seen https://github.com/willswang/vlc ?
[16:38] <av500> somebody wrote vlc?
[16:39] <av500> ah cedar
[16:42] <cbsrobot> stabbing, forking and spooning - youth today
[16:42] <divVerent> I know what -gstabs does and what fork() does
[16:42] <divVerent> but what is a spoon
[16:42] <burek> there is no spoon...
[16:42] <divVerent> oh right
[16:48] <j-b> cbsrobot: of course not
[16:48] <j-b> cbsrobot: what is cedar?
[16:51] <av500> allwiner a10
[16:51] <av500> the video core
[16:52] <j-b> nice!
[16:54] <cbsrobot> spooning is the same as stabbing but just not with a sharp object
[16:55] <j-b> cbsrobot: this is quite interesting, indeed
[16:55] <cbsrobot> j-b: here you have some more info: http://linux-sunxi.org/CedarX
[16:56] <cbsrobot> j-b: the spooing is interesting ? yeah sure.
[16:56] <cbsrobot> *spooning
[16:56] <j-b> what about sporking?
[16:58] <cone-480> ffmpeg.git 03Stefano Sabatini 07cb309922cb19: doc/ffprobe: factorize definition of writers syntax
[16:58] <cone-480> ffmpeg.git 03Stefano Sabatini 07ee6ae8355e92: doc/filters: fix a few typos in the frei0r docs
[16:59] <durandal_1707> what about doing something?
[17:01] <cone-480> ffmpeg.git 03Michael Niedermayer 07c2cbc80ae9e3: ffv1enc: consider 2pass float rounding, fix loop
[17:09] <divVerent> j-b: sporking? it doesn't even have a knife in it too
[17:15] <durandal_1707> our next release should have number 17
[17:16] <j-b> divVerent: with a knife, it would be sporfing
[17:16] <saste> michaelni: regarding ticket #1731, are you aware of any reason for which the AVI muxer is ignoring the initial timestamp offset?
[17:16] <saste> should it be considered a bug, or a "feature"?
[17:18] <divVerent> j-b: sporfing, I like that word
[17:18] <divVerent> but it somewhat sounds like design by committee ;)
[17:18] <divVerent> wtf, this thing even exists on wikipedia
[17:19] <j-b> divVerent: I was not joking.
[17:20] <saste> durandal_1707, why 17?
[17:20] <j-b> spork, sporf and spife exist
[17:20] <divVerent> and chork... wtf
[17:57] <saste> michaelni, so the point is that avienc only take in consideration DTS, and ignores PTS
[17:57] <saste> so possibly there is no solution to the problem, is that correct?
[17:58] <saste> even in this case, it would be useful to mark this feature with a format capability, so the user knows what will and won't work with a given muxer
[18:22] <michaelni> saste, there are no PTS in avi
[18:23] <saste> michaelni, so am I correct to claim that there is no way to record an initial timestamp "offset" in AVI?
[18:23] <saste> second question, would you agree to add an option to mark the feature in the format descriptor?
[18:24] <saste> this way it would be possible to warn the user in case the timestamp information is going to be lost in the remuxing
[18:24] <saste> (and would be possible to select a muxer which can do the job)
[18:27] Action: saste wonders how AVI deals with B-frames if it ignores PTS
[18:29] <saste> also regarding the intial timestamp loss problem, a video timepad filter may be helpful even in that case
[18:30] <michaelni> saste, "start time" should be possible in avi but maybe we talk about different things
[18:30] <saste> michaelni, ffmpeg -f lavfi -i "testsrc=s=qcif:d=1, setpts=PTS+5/TB,showinfo" -y out.avi
[18:32] <michaelni> saste theres a field in avi for the initial TS offset per stream 
[18:32] <saste> but, is it honored?
[18:35] <michaelni> our demuxer honors it
[19:27] <cone-480> ffmpeg.git 03Alexander Strasser 077750c48d3040: dshow: Use NO_DSHOW_STRSAFE for all dshow header inclusions
[21:37] <cone-480> ffmpeg.git 03Michael Niedermayer 072207ea44fb4f: ff_emulated_edge_mc: fix integer anomalies, fix out of array reads
[21:37] <cone-480> ffmpeg.git 03Michael Niedermayer 076535d81d8788: g723_1dec: Fix lsp2lpc() so it can handle values at the ends of the table.
[22:00] <Compn> j-b : what do you do about sites redist vlc or ffmpeg with spyware installer ? like http://www.xvidupdate.com/
[22:02] <ubitux> he posts a ragecomic on 4chan to call for ddos
[22:05] <Compn> haha
[22:27] <michaelni> Compn, this sounds like a question you should ask a lawyer (maybe SFLC)
[22:28] <j-b> Compn: DMCA
[23:17] <cone-480> ffmpeg.git 03Clément BSsch 07375cd3f2ecda: pp: rework the way templating is done.
[23:17] <cone-480> ffmpeg.git 03Clément BSsch 07c6945228e8eb: pp: simplify postProcess().
[00:00] --- Sat Nov 17 2012