[FFmpeg-devel-irc] IRC log for 2010-09-08

[irc at mansr.com]

[04:30:10] <Dark_Shikari> http://whatthecommit.com/
[05:20:19] <thresh> old news.
[05:21:00] <thresh> and moroning, fellow ffmpeg developers
[05:22:06] <funman> moroning, david
[05:29:53] <kshishkov> thresh, it's wonderful Swedish morning here
[05:31:52] <thresh> kshishkov: did I say I hate you?
[05:33:09] <kshishkov> thresh: nope, but you'd hate anybody saying similar thing. KotH, for instance
[05:33:17] <thresh> true
[05:36:14] <pJok> god morgon, kshishkov :)
[05:37:38] <kshishkov> goda morgnar, pJok
[05:37:59] <pJok> hur är läget?
[05:38:04] <kshishkov> thresh: you know, relocating somewhere else solves some problems
[05:38:32] <kshishkov> pJok: lagom bra
[05:38:57] <pJok> enjoying stockholm?
[05:39:29] <kshishkov> of course!
[05:41:58] <thresh> kshishkov: yeah I know
[05:43:05] <kshishkov> thresh: well, it finally worked for me - Ukrainian guy without good working experience and such
[05:45:04] <thresh> kshishkov: at least you have a degree
[05:46:07] <kshishkov> thresh: only a formal one, in unhuman area
[05:47:07] <_av500_> god moroning
[05:52:31] <CIA-11> ffmpeg: reimar * r25073 /trunk/libavcodec/vp3.c:
[05:52:31] <CIA-11> ffmpeg: Only draw the actually visible area in vp3_draw_horiz_band.
[05:52:31] <CIA-11> ffmpeg: Fixes a black line in non-swapped, non-mod-16-height Theora videos
[05:52:31] <CIA-11> ffmpeg: when vp3_draw_horiz_band is used.
[06:31:33] <KotH> a wonderfull swiss morning to everyone, especially thresh!
[06:31:55] * pJok sees no swiss mornings around here
[06:32:42] <KotH> then you should consider moving somewhere, where you can enjoy fresh, and beautifull swiss mornings!
[06:33:45] <pJok> i live in sweden, i have wonderful, fresh and beautiful swedish morrnings ;)
[06:34:38] <KotH> i wouldnt be able to stand swedish mornings... for the lack of swiss chocolate ;)
[06:36:25] <pJok> there's plenty of swiss chocolate in sweden
[06:38:07] <KotH> which ones? lindt? sprüngli? teuscher? callier? maestrani? munz?
[06:38:54] <KotH> and do you get lindt's chocoloit froid?
[06:39:12] <KotH> er.. chocolat froid
[06:39:32] <KotH> s/lindt/sprüngli/
[06:39:50] * KotH seems to not have head enough turkish tea this morning
[06:39:57] <KotH> s/head/had/
[06:39:59] <KotH> ^^'
[06:42:26] <lu_zero> good morning!
[06:42:34] <mru> moroning
[06:43:24] <KotH> bon giorno lu_zero
[09:45:07] <spaam> mru: you dont like tidsp ? :)
[09:45:32] <mru> I don't like felipe
[09:45:44] <mru> nobody does
[10:03:33] <CIA-11> ffmpeg: stefano * r25074 /trunk/libavcodec/ppc/ (6 files):
[10:03:33] <CIA-11> ffmpeg: Merge has_altivec() function into mm_support(), remove it and use
[10:03:33] <CIA-11> ffmpeg: mm_support() instead.
[10:03:33] <CIA-11> ffmpeg: Reduce complexity and simplify pending move to libavutil.
[10:05:20] <superdump> mru: not even the gstreamer peoples like him :)
[10:05:46] <mru> but they don't like sensible people either
[10:09:36] <lu_zero> what did he do?
[10:09:53] <mru> right now he's telling lies on ffmpeg-devel
[10:22:43] <lu_zero> tell me more =)
[10:24:40] <mru> why don't you read the ml?
[10:31:51] <funman> tl;dr?
[10:32:47] <funman> did Felipe Contreras write this 'tidsp' patch for his own or for a company?
[10:33:40] <av500> own
[10:34:35] <funman> crazy guy
[10:35:22] <funman> working with the ti dsp API was one of my most confusing coding experience
[10:35:52] <funman> that would be 'one of the dozens of APIs' it seems
[10:37:48] <av500> which one?
[10:38:41] <av500> you did that neuros stuff, right?
[10:39:19] <funman> yes
[10:39:29] <av500> so CE/dsplink
[10:39:34] <funman> that's the one yes
[10:39:41] <av500> yes, I fought that too for 3ys :)
[10:40:19] <funman> neuros project failed but some other people wanted to use the code
[10:40:30] <av500> funman: yes I know
[10:40:47] <thresh> funman: failed because of ?
[10:40:54] <av500> osd2 is still ok as a dm6446 evm...
[10:41:06] <av500> thresh: because they promised hd and never got it
[10:41:09] <funman> thresh: too high requirements
[10:41:18] <funman> the board was meant to decode NTSC/PAL
[10:41:23] <av500> and because they thought it could be done without writing code
[10:41:38] <thresh> ah I though we could blame funman
[10:41:41] <funman> adn they wanted 1080 capture + encoding -> stream + display
[10:41:47] <av500> 1080p lol
[10:43:00] <funman> also with me smoking dutch weed, no chance it could happen ;)
[10:59:17] <funman> sorry, it seems it was 720, not 1080
[11:05:50] <cartman> libmpcodecs/vf_yadif.c:262:9: error: ambiguous instructions require an explicit suffix (could be 'cmpb', 'cmpw', 'cmpl',
[11:05:54] <cartman>       or 'cmpq')
[11:05:59] <cartman> somebody needs to tell michael
[11:07:03] * funman thinks somebody should be cartman
[11:07:46] <cartman> funman: he would flame me to hell
[11:07:47] <cartman> :P
[11:08:01] <mru> cartman: aren't you going there anyway?
[11:08:51] <funman> michael is not a big flamer i guess
[11:08:53] <cartman> mru: a special hell my friend, no michael's hell
[11:09:20] <cartman>   "_MM_FIX_1_414213562", referenced from bla bla not found
[11:09:31] <cartman> how do I find out what _MM_FIX is?
[11:09:33] <cartman> looks mangles
[11:09:35] <cartman> mangled*
[11:09:35] <mru> gre
[11:09:36] <mru> p
[11:09:45] <cartman> oh not mangled?
[11:10:04] <mru> I guess the real name is MM_FIX_1_414213562
[11:10:14] <mru> that's sqrt(2) btw
[11:10:19] <cartman> oh there it is
[11:10:37] <funman> i'd swear sqrt(2) is somewhere between 1 and 2
[11:10:46] <funman> not over 4 billions!
[11:10:55] <mru> 1.414... is between 1 and 2
[11:11:40] <mru> FIX implies it's a fixed-point number
[11:11:50] <mru> so the point (.) has been fixed to a _
[11:12:04] <cartman> looks like an mmx define problem
[11:12:05] <funman> Qsomething.9 ?
[11:12:13] <cartman> wonders of mplayer
[11:12:29] <funman> i think the '.9' would be 9 bits not 9 digits
[11:12:47] <mru> bits are digits
[11:12:59] <mru> in the one true base
[11:13:17] <mru> or is that the two true base?
[11:13:21] <mru> base-1 is boring
[11:13:26] <funman> digit -> finger
[11:13:32] <mru> romans used base-1 and look what good it did for them
[11:13:39] <mru> and xiph
[11:13:58] <mru> how many fingers does your cpu have?
[11:13:59] <funman> the too true base
[11:14:07] <mru> true to the base
[11:14:07] <funman> one per ARM!
[11:15:11] <mru> the thumb is not a finger
[11:15:15] <mru> although it is a digit
[11:15:17] <cartman> _MM_FIX_1_082392200 and _MM_FIX_1_414213562
[11:15:20] <cartman> only those fail
[11:15:22] <cartman> weird
[11:15:26] <cartman> they are defined indeed
[11:15:44] <funman> mru: source?
[11:15:59] <mru> any dictionary
[11:16:30] <funman> http://en.wiktionary.org/wiki/digit doesn't seem to agree
[11:16:40] <mru> try a real dictionary
[11:16:49] <mru> OED or webster's would be good choices
[11:17:01] <cartman> mru: can I force fate to send the last report? ssh timed out
[11:17:17] <mru> you can do it manually
[11:17:27] <cartman> mru: could you tell me how?
[11:17:38] <funman> http://www.wordreference.com/enfr/digit ?
[11:17:42] <thresh> hmm, I believe subtitles in mkv are broken in 24911 at least and not in 24504. The subtitle stays on a screen waiting for next one to appear
[11:17:46] <funman> i can never find the free oed
[11:17:53] <thresh> how do I debug that? I have a behaviour like that in xbmc and vlc
[11:18:09] <funman> thresh: git bisect?
[11:18:13] <mru> cartman: sorry, it's probably deleted the files already
[11:18:21] <mru> maybe I should change that
[11:18:26] <cartman> mru: alrighty
[11:18:42] <mru> or maybe not
[11:18:57] <funman> that would need 9 compilation
[11:18:58] <mru> but never mind, it'll soon run again
[11:21:57] <thresh> funman: and vlc recompilation as well
[11:22:16] <cartman> Isn't this weird: http://pastebin.com/raw.php?i=UbzCZfvd
[11:22:23] <funman> http://www.oxforddictionaries.com/view/entry/m_en_gb0225580#m_en_gb0225580 < a finger or thumb
[11:23:18] <funman> cartman: this is x86?
[11:23:27] <cartman> funman: x86-64
[11:36:53] * lu_zero re-reads the thread
[11:37:05] <lu_zero> oh new flame in the end, I was missing that part
[11:37:23] <lu_zero> for me he was just a prick
[13:04:28] <lu_zero> uhm
[13:05:34] <lu_zero> the bitstream filters kick in the wrong time =|
[13:07:35] <lu_zero> or sdp kicks in the wrong time
[13:07:37] <lu_zero> =P
[13:11:54] <lu_zero> wbs: ping
[13:13:54] <wbs> lu_zero: pong
[13:14:40] <lu_zero> I'm trying to figure out how to fix the aac adts
[13:15:01] <lu_zero> I fixed it in feng already in a _quite_ gory way
[13:19:14] <lu_zero> the simplest/sanest way would be having the bsf kick in on probe
[13:19:46] <wbs> umm.. so what are you trying to do with lavf then? decode such a stream from feng?
[13:21:05] <lu_zero> with feng I do something gory as in
[13:21:32] <lu_zero> pick the first aac packet and feed it to the bsf on init
[13:21:45] <lu_zero> and then make the bsf work as in ffmpeg
[13:23:36] <wbs> don't know aac/adts well enough to know exactly what the bsf does for it
[13:24:17] <lu_zero> basically chops the first 2 bytes
[13:24:22] <lu_zero> more or less
[13:28:40] <lu_zero> we already call av_read_frame_internal on av_find_stream_info
[13:28:58] <lu_zero> so we would be fine if we feed the bsf that time
[13:41:05] <lu_zero> hi BBB
[13:41:29] <lu_zero> do you have more or less knowledge of what happens in av_find_stream_info ^^?
[13:42:05] <wbs> lu_zero: I don't know much about that at least, I'd have to go RTFS to be able to even give guesses ;P
[13:42:46] <lu_zero> it calls the parser
[13:42:58] <lu_zero> but we cannot abuse it I'm afraid
[13:59:05] <BBB> wbs: how do you like your first audio codec?
[13:59:15] <BBB> wbs: want to start doing h264 already?
[14:00:25] <BBB> maybe I should review that code too...
[14:00:28] * BBB busy :(
[14:00:54] <mru> working xvp8 yet?
[14:01:15] <BBB> starting it again...
[14:01:20] <BBB> h264 is done now
[14:01:29] <BBB> at least the stuff I intended to do
[14:04:33] <wbs> BBB: it's quite interesting, although it isn't "mine" really, most of the work was done by Kenan Gillet and all the previous authors
[14:04:47] <BBB> getting it in svn is 50% of the work
[14:04:50] <BBB> as you've noticed :-p
[14:04:56] <wbs> yes, I've noticed. :-)
[14:05:09] <wbs> hopefully it's good enough soon, so I can start working on the encoder
[14:05:32] <wbs> (michael rejected the encoder earlier saying that it needs some kind of trellis to test more than one alternative)
[14:05:48] <wbs> but I don't want to start on that one (since they share lots of code) until the decoder is settled
[14:08:50] <BBB> makes a lot of sense
[14:19:11] <DonDiego> stefano broke ppc
[14:20:41] <mru> which commit?
[14:20:52] <DonDiego> has_altivec removal was not complete
[14:21:01] <DonDiego> there are leftovers in libxvidff.c
[14:21:09] <mru> feel free to fix it
[14:21:33] <lu_zero> meh
[14:21:40] <mru> and wtf is that doing there?
[14:22:05] <DonDiego> who reviewed his commit?
[14:22:12] <DonDiego> there are multiple issues..
[14:22:20] <mru> michael did
[14:22:45] <mru> what else is wrong?
[14:24:58] <DonDiego> HAVE_ALTIVEC #ifdef addition
[14:25:08] <DonDiego> the file is compiled conditional to HAVE_ALTIVEC
[14:26:02] <mru> you could just fix it you know...
[14:26:57] <DonDiego> my fix is compiling..
[14:27:32] <DonDiego> what? i'm not allowed to rant in between? i thought this was ffmpeg ;)
[14:28:01] <mru> how are you fixing it?
[14:28:24] <mru> I say compile the file unconditionally
[14:29:00] <mru> the intent seems to be for that function to become generally available
[14:30:01] <BBB> there's four implementations of the function, no?
[14:30:07] <BBB> if HAVE_ALTIVEC, that one is used
[14:30:14] <BBB> otherwise another one is used (e..g in x86_cpu.h)
[14:30:17] <BBB> etc
[14:30:43] <mru> eventually we'll have to make sure there always one available
[14:30:56] <mru> now it only exists for some architectures
[14:31:13] <mru> and ppc w/o altivec certainly shouldn't use the x86 one
[14:31:29] <BBB> there was a generic one alwasy returning 0 in his patch
[14:31:32] <BBB> I think, at least
[14:31:58] <BBB> I think ppc w/o altivec uses the return 0; one
[14:32:01] <BBB> just like x86 w/o mmx does
[14:32:13] <mru> lots of hacks
[14:32:18] <mru> not the way to go
[14:33:04] <mru> for the architectures that have one, it should always be used
[14:33:30] <BBB> I thought !HAVE_MMX meant "I don't want optimizations"?
[14:33:36] <BBB> at least in code that's what it means
[14:33:41] <BBB> same for HAVE_ALTIVEC
[14:33:50] <DonDiego> mru: btw, can you please add 'make checkheaders' to the list of fate tests?
[14:33:52] <mru> but that should be hidden in the ppc dir
[14:33:54] <BBB> it's just that the altivec tests are buggy and poor, so it often returns 1 unconiditionally
[14:34:02] <mru> or the x86 dir as the case may be
[14:34:11] <BBB> right, I think that is the case now
[14:34:14] <BBB> I might be wrong though
[14:34:26] <mru> the generic code should check for PPC, not altivec
[14:35:44] <DonDiego> of course, 'make checkheaders' is currently broken
[14:36:00] <DonDiego> and it will remain broken forever unless the test is turned on in fate..
[14:36:10] <mru> nag, nag, nag
[14:37:19] <merbzt> oh and the warning counter
[14:37:19] <CIA-11> ffmpeg: stefano * r25075 /trunk/libavcodec/libxvidff.c:
[14:37:19] <CIA-11> ffmpeg: Remove reference to not anymore existing symbol has_altivec and use
[14:37:19] <CIA-11> ffmpeg: mm_support instead.
[14:37:19] <CIA-11> ffmpeg: Fix compilation if altivec is present and libxvidff.c is compiled.
[14:37:22] <merbzt> and a pony
[14:37:46] <lu_zero> uhmm
[14:37:53] <mru> merbzt: what purpose does a warning counter serve?
[14:39:12] <DonDiego> count warnings, what else?
[14:39:14] <DonDiego> :)
[14:39:30] <mru> sure, but what will the number tell you?
[14:39:53] <lu_zero> wbs: the applehttp_demuxer would work already on files?
[14:41:07] <wbs> lu_zero: if you have a .m3u8 locally, it should work just fine, I think
[14:41:23] <merbzt> mru: well the idea is that we shouldn't add code that adds warnings, and the warnings we have we should get rid of
[14:41:37] <mru> merbzt: that's a battle you cannot win
[14:41:41] <merbzt> thus a trend graph might motivate that
[14:41:47] <mru> every commit by michael adds warnings
[14:41:50] <merbzt> mru: I know
[14:41:53] <mru> and he refuses to fix them
[14:42:00] <mru> and he won't let anyone else fix them
[14:42:57] <lu_zero> wbs: picking the files from the hd?
[14:43:12] * lu_zero is reading the sourcecode before trying
[14:43:34] <lu_zero> (you can guess what I'm thinking to do)
[14:43:36] <merbzt> anyway if you are against it I wont bring it up any more, I just consider it a somewhat interesting metric
[14:43:50] <merbzt> ie the warnings over time went down or not
[14:44:24] <mru> I'm not against it as such
[14:44:30] <mru> I simply don't see it as very useful
[14:44:40] <mru> and hence not very high priority
[14:44:42] <wbs> lu_zero: well, it reads them from whatever url the .m3u8 file specifies... if it contains absolute urls, it reads them, if it contains relative urls, it applies them relative to the original file path
[14:45:40] <lu_zero> that's how I understood it ^^
[14:46:20] <wbs> hmmm, make_absolute_url might not handle cases where the original path is a simple filename (without an absolute path) though
[14:46:30] <wbs> but it should be quite simple to fix I think
[14:47:22] <lu_zero> could you point me the line?
[14:47:36] <wbs> applehttp.c:99-101
[14:48:30] <lu_zero> thank you
[14:48:48] <wbs> I think it could be fixed with else buf[0] = '\0';, that is, if the base url doesn't contain any slash at all, just replace it with the new url that should be relative to it
[14:49:09] <DonDiego> mru, merbzt, i support the idea of a warning counter
[14:49:37] <mru> I think our time is better spent on things that might make a difference
[14:49:57] <lu_zero> file:// should do the trick I guess
[14:50:38] <wbs> lu_zero: yes, or just an absolute path to begin with should work just fine
[14:51:02] <lu_zero> uhmm
[14:51:27] <lu_zero> ffmpeg -i my.m3u8
[14:51:32] <lu_zero> my.m3u8: Not a directory
[14:51:41] <av500> which is true
[14:51:46] <lu_zero> ffmpeg usually helpful message
[14:53:10] <av500> its better than "internal error"
[14:53:17] <lu_zero> the file:// hack work
[14:53:30] <lu_zero> now I can check and see how to unhack it
[15:08:10] <CIA-11> ffmpeg: stefano * r25076 /trunk/ (36 files in 8 dirs):
[15:08:10] <CIA-11> ffmpeg: Move mm_support() from libavcodec to libavutil, make it a public
[15:08:10] <CIA-11> ffmpeg: function and rename it to av_get_cpu_flags().
[15:08:46] <lu_zero> ok there is a small annoyance
[15:09:04] <lu_zero> the dts appear to be non monotonous
[15:09:25] <lu_zero> (on the cut between the segments you have the same timestamp)
[15:11:29] <CIA-11> ffmpeg: stefano * r25077 /trunk/doc/APIchanges: Add APIchanges entry for the av_get_cpu_flags() addition of r25076.
[15:25:12] <kshishkov> wbs: today I saw cheese and thought of you.
[15:28:46] * av500 does not want to be remebered like that
[15:29:29] <kshishkov> av500: you're lucky, I remember you when I see certain video codec family
[15:30:03] <mru> rv?
[15:30:23] <kshishkov> no, from Off3
[15:31:57] <kshishkov> his .de site distracts a bit though
[15:32:27] <mru> vp7
[15:35:05] <CIA-11> ffmpeg: diego * r25078 /trunk/libavcodec/a64colors.h: Add missing stdint.h #include; necessary to pass 'make checkheaders'.
[15:35:26] <wbs> kshishkov: oh, what cheese was that? :-)
[15:37:06] <kshishkov> wbs: Storsjö, dunno why I thought about you or certain place in Osterbotten
[15:37:28] <wbs> kshishkov: oh, there's such a cheese too? scary :-)
[15:38:07] <kshishkov> wbs: scary is when they get cheese with my name
[15:38:13] * mru thinks of Storsjöodjuret
[15:40:36] <spaam> yeah..
[15:40:39] <spaam> same here :)
[15:41:55] <kshishkov> mru: and your name reminds me of certain Astrid Lindgren character
[15:42:11] <mru> which one?
[15:42:13] <kshishkov> Pippilotta Viktualia Rullgardina Krusmynta Efraimsdotter LÃ¥ngstrump
[15:42:25] * mru is not a fan of astrid lindgren
[15:42:26] <kshishkov> (should be obvious)
[15:42:41] <mru> her books are silly and childish even for children's books
[15:42:58] <kshishkov> you haven't read Soviet books then
[15:43:17] <spaam> kshishkov: can you read about stalin and lenin in those? :)
[15:43:28] <kshishkov> spaam: Lenin only
[15:43:42] <kshishkov> there were even series of tales about Lenin for children
[15:44:02] <spaam> so he was a great leader then :)
[15:44:05] <mru> kshishkov: those were just rebranded superman comics
[15:44:24] <av500> kshishkov: the ones with the cute little mole?
[15:44:30] <kshishkov> mru: no, they were written by some electronic guy who knew Lenin actually
[15:44:40] <av500> DJ Lenin?
[15:44:44] <kshishkov> av500: cartoons only
[15:46:22] <kshishkov> well, and all books told you how to behave, what ideals you should have, etc, etc
[15:47:31] <kshishkov> there was (probably true story) a book for children about firefighters, it started with words "Firefighting in Soviet Union is cardinally different from Tsarist Russia"
[15:48:54] <av500> it changed from "let it burn unless belongs to tsar" to "let it burn unless belongs to party"?
[15:49:42] <mru> "it will burn unless belongs to party"
[15:50:01] <kshishkov> "it will burn regardless but we'll justify it by party needs"
[15:50:03] <av500> it was about firefoghters, not firestarters...
[15:50:16] <mru> can't have one without the other
[15:52:36] <kshishkov> still it's a good start for children's book
[15:52:57] <mru> in soviet russia, fire fight you?
[15:53:06] <kshishkov> ask thresh
[15:59:07] <kshishkov> BTW,  anyone knows children book by Norwegian author full of trolling and troll logic?
[15:59:26] <kierank> mru will know
[15:59:31] <av500> mru wrote it
[15:59:42] * mru doesn't write norwegian
[15:59:49] <kierank> and av500 wrote the preface
[16:00:55] <kshishkov> damn, where are those Nordic sheiks when you need to ask them?
[16:01:31] <peloverde> lu_zero: Can you tell me more about the hack you use for h264?
[16:03:30] <CIA-11> ffmpeg: diego * r25079 /trunk/libavcodec/aac_tablegen_decl.h: Use filename as pattern for multiple inclusion guards.
[16:17:31] <DonDiego> mru: why is %_tablegen.h part of SKIPHEADERS in libavcodec?
[16:17:45] <DonDiego> it looks as though they should be a part of it
[16:17:55] <mru> "it"?
[16:18:25] <DonDiego> and then there is aac_tablegen_decl.h which is not matched by that pattern
[16:18:40] <DonDiego> it looks as though they should be a part of ALLHEADERS like any other header file
[16:18:50] <DonDiego> sorry, that first sentence was unclear
[16:19:03] <mru> they can't in general be compiled alone
[16:19:44] <DonDiego> the aac ones can..
[16:19:53] <mru> coincidence
[16:20:07] <mru> or bad naming
[16:20:14] <DonDiego> the latter i think
[16:20:28] <DonDiego> and the _decl thing is even more unfortunate
[16:20:34] <DonDiego> peloverde: you around?
[16:20:48] <peloverde> yes
[16:22:16] <peloverde> the aac tablegen headers are annoying because some are used by both enc and dec
[16:22:55] <peloverde> See the thread :"[PATCH] aactab: Tablegenify ff_aac_pow2sf_tab."
[16:24:07] <DonDiego> well, fix the naming..
[16:25:37] <BBB> mru: re:sws_cpu, thanks fr stating the obvious
[16:25:56] * BBB still can't believe people think compile-time cpu detection should require a completely different codepath
[16:26:12] <DonDiego> oh, i see that the cyclic dependencies between avformat.h and avio.h have worsened..
[16:26:57] <kshishkov> mru: good political advertisement here - "a, b, c, d, e, fp", makes me think of some 8-bit CPU regs
[16:27:53] <DonDiego> peloverde: is it safe to add config.h #includes to aac_tablegen.h and aac_tablegen_decl.h?
[16:28:01] <mru> no
[16:28:30] <mru> those are built for the host when cross-compiling
[16:29:35] <DonDiego> then aac_tablegen_decl.h needs to be renamed or the pattern adapted
[16:31:37] <BBB> DonDiego: do you want to go to amsterdam?
[16:31:40] <BBB> to give a talk
[16:31:48] <DonDiego> when?
[16:31:50] <DonDiego> where?
[16:31:51] <DonDiego> what?
[16:32:00] <mru> why?
[16:32:03] <BBB> :-p
[16:32:06] <DonDiego> what topic?
[16:32:21] <DonDiego> but yes, i'd sure enjoy going there and giving a talk :)
[16:32:53] * mru can come along to troll him
[16:33:26] <DonDiego> oh, a private troll :)
[16:33:54] <mru> it's one of the services I can provide
[16:33:56] <kshishkov> nope, comunity one
[16:34:48] <BBB> fri sep 17 2010
[16:34:52] <BBB> in a'dam
[16:35:04] <av500> say "ken sent me"
[16:35:14] <kshishkov> give him a topic or grass too
[16:35:15] <BBB> they'd like a presentation by someone from ffmpeg about our video codecs and specifically vp8, technical is ok but it can be more global about our quest for world domination also
[16:35:29] <BBB> I'm not anywhere near a'dam so it's hard to go for me
[16:35:32] <av500> who is "they"?
[16:35:55] <kshishkov> av500: were you referencing certain leisure suit game?
[16:36:04] <av500> ocourse
[16:36:04] <DonDiego> i'm not the most technical guy around, but sure, i can give a presentation..
[16:36:15] <mru> you're near amsterdam
[16:36:20] <BBB> you're a good talker from what I've heard
[16:36:28] <BBB> and conveniently located indeed
[16:36:39] <BBB> and you can ask yuvi and me to help with technical bits
[16:36:40] <BBB> ;)
[16:36:47] <mru> fwiw, I can't make it that friday
[16:36:59] <mru> working office hours sucks
[16:37:46] <kshishkov> mru: working usually sucks, office hours or not
[16:38:04] <mru> the work itself is ok
[16:38:11] <mru> you know what I'm doing
[16:38:41] <DonDiego> BBB: ok, i'm on it, but who am i supposed to talk to?
[16:38:53] <DonDiego> train fare to amsterdam is 33eur
[16:39:06] <kshishkov> mru: yes, feeling bad since the guy who wrote those bad interfaces is also on vacation
[16:39:25] <DonDiego> damn deutsche bahn, i can barely go half that distance at the same price in .de..
[16:39:28] <mru> the interfaces could be worse
[16:39:41] <mru> DB is expensive
[16:40:17] <DonDiego> bbiab
[16:40:23] <kshishkov> at least you can have DB cards
[16:40:37] <mru> sucks if you're only visiting
[16:40:43] <av500> and free food in 1st
[16:41:04] <kshishkov> av500: _you_ call it food?
[16:41:21] <av500> i know, in soviet russia, food calls you....
[16:41:58] <kshishkov> no, you could say "free snack" or "free edible particle"
[16:42:20] <mru> free non-toxic fragment
[16:42:34] <av500> ffntf ftw
[16:44:17] <kshishkov> at least 2nd class in SJ much better than on DB
[16:44:31] <lu_zero> btw
[16:44:57] <kshishkov> yes, Italian trains are the worst
[16:44:59] <lu_zero> mru: you have your previous lt presentations around?
[16:45:26] <mru> lu_zero: never given any
[16:45:31] <lu_zero> I'd put all of them on the foundation website
[16:45:52] <lu_zero> weren't you preparing something 2 years ago at LT?
[16:46:00] <mru> not me
[16:46:01] <kshishkov> lu_zero: av500 has it, it's called "the beast". Youtube link should be enough
[16:46:06] <BBB> woops
[16:46:08] <BBB> had to run out
[16:46:11] <lu_zero> kshishkov: =P
[16:46:30] <BBB> DonDiego: I'll forward you the email I received so you can talk to the guy about what/how and so on, he should reimburse you for travel expenses etc.
[16:46:31] <kshishkov> lu_zero: or ask Diego, he's our more humanistic guy
[16:46:35] <av500> kshishkov: I can also give beast to DonDiego but he cannot carry it
[16:47:06] <lu_zero> BBB: btw which is the situation with the eu conservancy?
[16:48:20] <BBB> lu_zero: I'll ask aaron
[16:48:22] <janneg> lu_zero: mike gave a presentation at lt
[16:49:04] <kshishkov> yes, slides should be somewhere in his blog
[16:50:32] <janneg> http://www.linuxtag.org/2007/de/conf/events/vp-speakers/vortragsdetails-talkid-87.html
[16:50:42] <Dark_Shikari> BBB: so, how's xvp8 going?
[16:50:55] <BBB> damnit :-p
[16:51:10] <BBB> I'm actually on that again
[16:51:18] <BBB> but am at work atm :-p
[16:51:31] <BBB> but I won't waste more time on h264_idct except for svn commit in 2-3 days
[16:53:28] <lu_zero> BBB: if needed I could put them in contact with some friendly lawyers of this side
[16:55:46] <BBB> ok, I'll tell him
[16:55:51] <BBB> are they free and non-profit? :-p
[16:57:02] <kshishkov> lawyers? non-profit? reread definition of "lawyer"
[16:57:23] <lu_zero> BBB: pro-bono possibly
[16:57:31] <lu_zero> I still need to chat with him
[16:58:10] <lu_zero> but he is quite focused on opensource
[17:03:47] <janneg> mru: there's at least a third one in libpostproc
[17:04:46] <kshishkov> janneg: we don't use postproc
[17:05:28] <janneg> mythtv does optionally
[17:05:57] <kshishkov> and some player hack
[17:06:05] <kshishkov> but not FFmpeg itself
[17:06:36] * av500 might use libpostproccore
[17:07:15] <kierank> BBB: who wants ffmpeg to give a talk?
[17:07:26] <kshishkov> av500: ok, go on and improve it then
[17:07:27] <janneg> what about libpostproccoreutil?
[17:07:42] <av500> hmm
[17:08:00] <av500> maybe the core bits of it
[17:08:42] <janneg> not needed, I'll do the same fix for libpostproc as for libswscale once we found an acceptable approach for swscale
[17:09:04] <kshishkov> full rewrite FTW!
[17:09:57] <lu_zero> ^^;
[17:10:01] <janneg> av500: btw, the demod part is listed by trident. They even have a formular to request datasheets
[17:10:18] <janneg> which I did, but no abswer yet
[17:10:30] <janneg> answer even
[17:11:06] <av500> ah, so trident was correct :)
[17:13:38] <BBB> kierank: university of amsterdam
[17:13:40] <janneg> yes, I know why I've missed the news. the heading was "micronas refocuses on automotive chips"
[17:19:59] <kierank> [17:44] <@kshishkov> yes, Italian trains are the worst --> italian trains are good
[17:20:07] <kierank> except there isn't much train safety
[17:20:28] <kierank> i.e. passengers crossing the tracks at the station
[17:22:15] <kshishkov> kierank: but no Italian said anything against my words, so it must be true
[17:22:51] <kierank> trains anywhere > british trains
[17:23:30] <kshishkov> kierank: that's bold statement, .su/.ru/.ua trains are even worse
[17:23:38] <kshishkov> at least they should be
[17:23:48] <av500> kierank: do they still match the minimum definition of a train "service" at all?
[17:24:07] <kierank> doubtful
[17:25:04] <kshishkov> av500: what's the definition?
[17:25:15] <av500> swiss trains :)
[17:25:16] <kierank> must arrive
[17:26:27] <kshishkov> av500: i.e. train service is trains running in mountain area with announcements on outdated dialects?
[17:26:50] <av500> and "on time" :)
[17:27:04] <kierank> av500: not in uk. presumably we have an eu exclusion for that
[17:27:51] <kshishkov> well, in .ua they announce when train is delayed by more than 15 minutes
[17:28:12] <mru> in .se they announce when a train is on time
[17:28:55] <kshishkov> what train?
[17:29:34] <kierank> I think +- 20 minutes is still on time here
[17:30:43] <kshishkov> well, Stephenson is long dead
[17:31:43] <av500> the patent troll?
[17:31:52] <kierank> it's his fault our gauge is different
[17:32:25] <CIA-11> libswscale: ramiro * r32068 /trunk/libswscale/ (swscale.h rgb2rgb.c options.c): rgb2rgb: build SSE2 codepath
[17:32:39] <kshishkov> he may be the last guy to service trains as well
[17:33:38] <av500> he took the oil can to his grave
[17:34:35] * kshishkov remembers typical British superhero - bicycle repairman
[17:35:53] <mru> kierank: having a different gauge isn't so bad on an island
[17:36:23] <kshishkov> same for road sides (unless you want to export or import cars)
[17:36:40] <mru> visitors renting cars get a hard time
[17:37:02] <kierank> mru: we can't have double decker trains because of the gauge
[17:37:03] <av500> kshishkov: with some more non-std things they could have protected the local car industry much better
[17:37:07] <av500> like 3 wheels
[17:37:15] <mru> kierank: too narrow?
[17:37:24] <kierank> yes something like that
[17:37:30] <kshishkov> kierank: why would you like double-deckers on the Isle?
[17:37:30] <av500> kierank: it could be sleeper trains
[17:37:44] <mru> av500: nothing is that far away
[17:37:52] <mru> it's a bloody island
[17:38:04] <kshishkov> av500: they did, and so did Soviets
[17:38:21] <av500> lying down, the height can be reduced a lot
[17:39:20] <mru> special trains for politicians, they lie a lot
[17:39:28] <kshishkov> av500: coffins on rails? Sounds good for Romania
[17:39:57] <av500> triple decker concept exists: http://www.instablogsimages.com/images/2008/08/19/07_MpBIQ_16269.jpg
[17:40:25] <mru> those things are real
[17:40:28] <BBB> Dark_Shikari: ping
[17:40:29] <av500> iknow
[17:40:34] <kshishkov> av500: I'll try it when it's comfortable for you
[17:40:39] <Dark_Shikari> BBB: pong
[17:40:50] <BBB> Dark_Shikari: can you commit the relicensing of h264_idct_sse2.asm?
[17:40:53] <av500> kshishkov: note the front part for them oversized people :)
[17:40:58] <BBB> (private email request by carl eugen)
[17:41:05] <BBB> Dark_Shikari: or pengvado is fine also
[17:41:43] <Dark_Shikari> ?
[17:41:44] <Dark_Shikari> why
[17:41:47] <kshishkov> av500: that's for not-undersized
[17:43:24] <av500> kierank: here's narrow gauge for you: http://farm1.static.flickr.com/77/187947975_98c0a680e6_o.jpg
[17:44:12] * kshishkov tried Roslagsbanan
[17:48:57] <mru> a railway in sweden: http://hardwarebug.org/files/dsc01197.jpg
[17:49:27] <spaam> nice mru
[17:50:16] <kshishkov> looks more like old mining railway
[17:51:39] <mru> it's an old quarry near älvdalen
[17:52:08] <kshishkov> of course, that region is known mostly for copper mines
[17:52:20] <mru> no, that's Falun
[17:53:35] <mru> Älvdalen is more known for porphyry
[17:54:28] <kshishkov> is it also known for Dalaåsna instead of Dalahäst?
[17:54:41] <mru> not that I know
[17:54:51] <mru> and the dalahäst is from Mora
[17:55:25] <kshishkov> ~40km from Älvdalen
[17:55:41] <mru> do not let the distance fool you
[17:55:47] <mru> it's a completely different place
[17:56:30] <kshishkov> maybe I'll visit it one day
[17:59:21] <mru> careful, my sister lives there
[17:59:29] <mru> and my mother not far off
[17:59:51] <kshishkov> they don't know how I look like
[18:00:03] <mru> I could tell them
[18:00:06] <spaam> scary.. family rullgård..
[18:00:13] <mru> spaam: you have no idea
[18:00:22] <kshishkov> spaam: and pet llama
[18:00:29] <spaam> kshishkov: mm
[18:00:37] <mru> and I'm leaving that topic now
[18:00:43] <kierank> mru's family lives under a bridge
[18:00:58] <mru> kierank: that I'd be able to handle
[18:01:05] <kshishkov> ok, Mora/Borlänge/Falun then
[18:02:06] <kshishkov> mru: I also have scary number of relatives at countryside. I'm extremely glad I've not seen most of them nor ever shall.
[18:10:53] <lu_zero> BBB: ping
[18:11:30] <lu_zero> Marco Ciurcina offers his full support to our cause =)
[18:11:50] <lu_zero> so as next step he needs to know better what we want ^^
[18:12:51] <lu_zero> (he's a good friend of Moglen and if he could help he'd be glad to)
[18:41:03] <BBB> lu_zero: awesome
[18:41:06] <BBB> I'll contact him, thanks
[18:41:16] <BBB> (and same for our privchat in my other window :-p)
[18:43:23] <lu_zero> email on the ml to discuss our needs first =P
[18:49:58] <BBB> ok
[19:36:33] <kierank> could av_read_frame be made to output multiple packets for containers with multiple packets per frame
[19:37:20] <mru> not while michael is alive
[19:39:31] <bcoudurier> hi guys
[19:50:29] <DonDiego> welcome frog eater :)
[19:59:22] <lu_zero> DonDiego: which one?
[20:00:18] <DonDiego> baptiste..
[20:14:11] <_av500_> kierank: so declare the packets to be subpackets :)
[20:19:50] <CIA-11> ffmpeg: bcoudurier * r25080 /trunk/libavformat/mov.c: In mov demuxer, do not override aspect ratio in tkhd by pasp like quicktime, fix issue #1539
[20:21:56] <CIA-11> ffmpeg: bcoudurier * r25081 /trunk/libavformat/mov.c: Check dref size based on a patch by google
[20:22:49] <bcoudurier> DonDiego :)
[20:23:30] <CIA-11> ffmpeg: bcoudurier * r25082 /trunk/libavformat/movenc.c: In mov muxer, write reduced sample aspect ratio values in pasp
[22:30:13] <janneg> bah, cherry-picking 25k commits takes a while
[22:31:00] <mru> manually?
[22:31:06] <mru> that would take a while...
[22:31:32] <mru> if you're automating it, do it in a scratch repo in a tmpfs
[22:31:43] <mru> much, much faster
[22:33:47] <janneg> automatically. I'm interleaving ffmpeg's history with libswscale's. working on patches touch libswscale is a pain otherwise
[22:35:36] <mru> it's not so bad if they are confined to libswscale only
[22:38:21] <janneg> unfortunately the are touching both
[22:40:06] <janneg> 23 minutes on a slow notebook harddrive
[22:40:50] <mru> tmpfs is like 100x faster
[22:41:17] <mru> because fsync is free there
[22:44:17] <spaam> hmm.. maybe run your system from a tmpfs!
[22:44:49] <spaam> i think i saw a guide for that on gentoo's forum some years ago :)
[22:55:29] <lu_zero> spaam: we had a friendly fork that used that as security measure
[22:56:09] <lu_zero> (run everything on ram, shutdown on tampering, good luck picking up anything)
[22:56:35] <spaam> Nice :)
[22:57:02] <lu_zero> http://en.wikipedia.org/wiki/Tin_Hat_Linux
[23:11:43] <spaam> ok :)
[23:36:57] <CIA-11> ffmpeg: ramiro * r25083 /trunk/libavcore/imgutils.h: av_fill_image_linesizes -> av_image_fill_linesizes