[FFmpeg-devel-irc] IRC log for 2010-06-15

irc at mansr.com irc at mansr.com
Wed Jun 16 02:00:51 CEST 2010


[00:14:10] <Dark_Shikari> mru: ping
[00:14:25] <Dark_Shikari> or Yuvie
[00:14:34] <Yuvie> ?
[00:14:55] <Dark_Shikari> question about SAD on NEON
[00:14:57] <Dark_Shikari> how fast is it?
[00:15:05] <Dark_Shikari> like, per 16 pixel row.
[00:17:27] <Yuvie> 6/2 to get 16 absolute differences
[00:17:55] <Dark_Shikari> so 16x16 -> 36 cycles?
[00:17:58] <Dark_Shikari> or so
[00:19:23] <Yuvie> err, 4/1 to get 16 absolute diffs, 2x 4/1 to get 16 absolute diffs + expand to 16 bit, 6/2 to add a new row of absolutediffs to existing sums
[00:19:38] <Dark_Shikari> so what's the total time for a 16x16 sad, about?
[00:22:28] <Yuvie> in theory: 36 for sad, +6+12 for horizontal sum, +20 for neon -> arm
[00:22:41] <Yuvie> benching x264's, 120 cycles
[00:23:46] <Dark_Shikari> so, I wrote in --me zero support to x264
[00:23:53] <Dark_Shikari> it helps like 20-30%+ without asm on x86
[00:24:03] <Dark_Shikari> it helps minimally, if not at all, with asm
[00:24:12] <Dark_Shikari> because motion search is so fast that it's actually worth it just to save time in cavlc.
[00:24:16] <Dark_Shikari> (on a core i7, at least)
[00:24:26] <Dark_Shikari> Yuvie: what's sad_x4_16x16 on neon?
[00:24:54] <Yuvie> 302 cycles
[00:25:05] <Dark_Shikari> heh.  ouch.
[00:25:15] <Dark_Shikari> I wonder why it's so high.
[01:01:18] <Dark_Shikari> also, mru, since you know more about non-x86 stuff, I'm curious what you think of that
[02:09:40] <Dark_Shikari> http://images.4chan.org/g/src/1276567615834.png for the gentoo users here
[02:09:41] <Dark_Shikari> ;)
[02:15:32] * Compn just glad that knoppix revolutionized linux installs
[04:08:49] <_av500_> knoppers ftw
[05:23:39] <barque> jesus
[05:23:42] <barque> who links 4chan
[05:23:43] <barque> seriously
[05:23:59] <barque> that entire site dumps information every 5 minutes
[05:24:09] <barque> at least the random board anyway
[05:24:30] <barque> and ummm Knoppix sucks.... just install Debian
[05:24:33] <barque> or Ubuntu
[05:24:43] <barque> Knoppix is only useful when all hell has broken loose
[05:24:50] <barque> it saves you
[05:32:36] <barque> oh god moot is an ass, don't visit /b/ right now
[05:48:33] <barque> the Uzbek/Kyrgyz ethnic clashes right now make me sick to my stomach
[05:48:37] <barque> hello Kosovo II
[05:48:45] <barque> *sigh* ... can't we all just get ... ALONG?
[06:18:45] * av500 waves to CIA-92
[06:22:55] <peloverde> CIA-92 doesn't like waves, it likes commits, submit more patches :)
[06:23:07] <barque> I'd help
[06:23:11] <barque> =)
[06:23:18] <barque> PLZPLZPLZPLZPICKMEEEEEEEEEEE
[06:23:23] <barque> IWANNASAVETHEWORLD
[06:23:42] <barque> how you folks doin' this awesome night?
[06:23:45] <av500> barque: fine, send a patch
[06:23:51] <barque> :D
[06:24:01] <barque> I'm gonna save some time, take a look at avfilter later
[06:24:20] <barque> I think adding some rgb,yuv command line modders is pretty slick
[06:24:50] <barque> whadya fink?
[06:24:56] <av500> send patch to ml
[06:25:05] <barque> right now I'm blazed out of my head listening to some sick dubstep
[06:25:10] <barque> http://www.youtube.com/watch?v=O_VFU4dKsr8&NR=1
[06:25:20] <barque> man this is good
[06:27:26] <barque> ffmpeg + lame == awesome... especially when I download mp3s extracted from flv streams
[06:27:38] <barque> these online rippers suck bawls
[06:35:13] <barque> ***I still haven't received my ffmpeg-devel confirmation e-mail***
[06:39:03] <barque> I registered with a different email too
[06:39:07] <barque> just right now
[06:39:25] <barque> nothing
[06:39:29] <barque> do you guys need fancy emails?
[06:39:39] <barque> I have a yahoo and a hotmail
[06:40:07] <merbzt> nope no fancy email
[06:49:46] <barque> umm,ok, if anyone here who is awake right now is in charge of mailing list I cam pm you my 2 email addresses
[06:49:50] <barque> so you can take a look
[06:49:58] <barque> I can*
[07:21:55] <kshishkov> barque: you know, nobody cares about Middle Asia, so should you
[07:22:26] <barque> lol?
[07:22:33] <barque> dobre den kshishkov
[07:22:38] <barque> kaktila? :]
[07:22:49] <barque> I'm actually from Canada
[07:23:35] * kshishkov is actually from Ukraine
[07:24:04] * elenril wouldn't admit that if he were kshishkov 
[07:24:29] <kshishkov> elenril: why?
[07:24:42] <barque> ahh I have a friend from Ukraine
[07:24:47] <barque> his name is Artur Oliva
[07:24:51] <barque> great guy
[07:24:55] <barque> also GSC and A4 games rule
[07:25:01] <elenril> kshishkov: shady commie background ;)
[07:25:11] <barque> I have all STALKERs plus Metro 2033
[07:25:23] * kshishkov preferred the times when GSC was releasing games from the West instead of creating own ones
[07:25:47] <barque> you guys at least make good games
[07:25:52] <barque> kharasho kharasho
[07:25:53] * elenril o_0 @ http://www.alsa-project.org/alsa-doc/alsa-lib/index.html
[07:26:03] <elenril> so alsa _can_ do dynamic configuration?
[07:26:11] <kshishkov> elenril: http://en.wikipedia.org/wiki/Yakov_Smirnoff - Ukrainians have an opinion on communism too
[07:26:24] <av500> elenril: I dont trust alsa further than I can throw it
[07:27:22] <elenril> av500: alternatives?
[07:27:38] <elenril> pulseaudio turned out to be even more evil, i just purged it
[07:27:55] <kshishkov> and OSS is a bit outdated :(
[07:28:18] <elenril> oss is a shady commie product
[07:28:24] <av500> elenril: I wonder why we have our own audiomixer....
[07:28:34] <elenril> av500: ?
[07:28:38] <av500> (at work)
[07:28:39] <barque> I thought we were all commies
[07:28:44] <barque> :(
[07:28:54] * barque goes home
[07:28:59] <kshishkov> well, USA will be with Obama
[07:29:03] <barque> LOL no way
[07:29:11] <barque> hint: read David Harvey
[07:29:29] <barque> USA right now is the biggest neoliberal reconfiguration system ever
[07:29:41] <elenril> av500: what's so wrong with alsa anyways
[07:30:12] <av500> elenril: it does not let me see the ioctls! :)
[07:30:16] <barque> neoliberal because they are implementing neoliberal policies (bailout)
[07:30:27] <barque> and reconfiguration because , wait until the next Reaganite comes along
[07:30:28] <elenril> av500: it doesn't?/why would you want to do that?
[07:30:49] <kshishkov> barque: that's socialism in action, they had it in Soviet Union too
[07:31:04] <barque> nah in the Soviet Union you had something called the Red October
[07:31:10] <av500> elenril: too feel in control :)
[07:31:19] <av500> wasnt that a submarine?
[07:31:21] <barque> if there is to be Socialism it will come in the form of a revolution of the exploited ploretrait
[07:31:46] <kshishkov> negroes coming to rulership?
[07:31:50] <barque> what will come (according to Marx) will be the dictatorship of the ploretrait
[07:31:55] <av500> kshishkov: omg
[07:32:24] <barque> which will evoke global class consciousness by setting an example and causing revolutions elsewhere
[07:32:49] <av500> yawn
[07:33:02] <barque> and after the ploretrait have made us accostumed to economic equality, the apparatus of the state will be abandoned and we will live in peaceful anarchy forever (according to both Marx and Engels)
[07:33:39] <barque> lol The End.
[07:33:40] <barque> hehehe
[07:33:55] * elenril thinks they were just trolling
[07:33:58] <jai> ploretrait?
[07:34:03] <jai> you mean proletariat
[07:34:10] <barque> yeah something along those lines
[07:34:16] <kshishkov> elenril: not as much as a guy called Lehnin
[07:34:24] <barque> the abundant mass of workers who are not artisans
[07:34:36] <kshishkov> proles for short
[07:34:51] <elenril> kshishkov: that's a different book
[07:35:18] <barque> who are exploited by the evil capitalists who use valorization of commodities and completely ignore labor-theory-of-value
[07:35:34] <av500> barque: yes, you took a lot of notes in school....
[07:35:35] <kshishkov> elenril: yes, British governmental plan 1984-...
[07:35:55] <barque> av500, lol I don't buy one Oz of this crock XD
[07:35:56] <av500> no need to reprint them here
[07:36:15] <barque> I'm a die hard libertarian (anarchist capitalist)
[07:36:19] <barque> XD
[07:36:22] <barque> bling bling baby
[07:36:40] <barque> $_$
[07:36:45] <barque> dolla dolla bill yo
[07:36:56] <barque> of course if you tie it down to gold or silver -_-
[07:37:09] * elenril goes to study some forbidden knowledge :(
[07:37:35] <kshishkov> Silvio Gesell?
[07:37:54] <barque> Murray Rothbard!
[07:37:56] <KotH> german?
[07:38:32] <barque> von Hayek
[07:38:34] <barque> Austrain baby
[07:38:38] <barque> Austrian*
[07:38:39] <barque> dammit
[07:38:46] <barque> Von Mises
[07:39:13] * kshishkov prefers Swedish economical school for some reason
[07:40:18] <elenril> meh, economics is a shamanism =p
[07:40:23] <barque> heh
[07:40:27] <barque> coding is way better XD
[07:40:38] <elenril> ForScience!
[07:40:55] <barque> FOR SCIENCE!
[07:41:08] <barque> of course, in all seriousness, I'm a huge proponent of sound science
[07:41:11] <kshishkov> elenril: partially. That tight Scottish guy did a lot of work making economics resembling science
[07:41:16] * elenril goes to do some research on the people who are still alive
[07:41:20] <barque> and science education
[07:41:39] <barque> Murray Rothbard was a big fan of hard science
[07:41:50] <barque> he did say that we shouldn't take ourselves VERY seriously in economics
[07:42:01] <barque> economics is like psychology
[07:42:07] <barque> macro of course
[07:42:47] <kshishkov> psychology is not a science at all
[07:43:00] <barque> now now, don't be so harsh
[07:43:02] <barque> it's soft science
[07:43:06] <barque> it tries to play with the big boys
[07:44:12] <kshishkov> and MS-DOS is a soft multitasking OS
[07:44:18] <barque> hahahah no
[07:44:21] <barque> lol
[07:44:34] <kshishkov> why? it can run several tasks at one time
[07:44:38] <barque> no it can't :D
[07:45:02] <ohsix> yes it can
[07:45:08] <barque> try not poping that 0 off of stack at the end see what happens
[07:45:11] <ohsix> called terminate and stay resident
[07:45:22] <ohsix> dig ralph browns interrupt list
[07:45:28] <barque> yeah you have all your ISRs dead
[07:45:30] <barque> in memory
[07:45:32] <barque> congrats
[07:46:40] <ohsix> kshishkov: if it weren't a science then there would be no theories and tests that validate them :]
[07:48:58] <barque> man good ol' days of 16 bit 286 real mode
[07:49:00] <ohsix> one aspect of psychology might explain why someone would be cross if you punched them in the face, is that something you can't  test and theorise around?
[07:50:03] <ohsix> barque: to be fair; dos proper would never run your stuff periodically; but tsr's were for software that chained dispatches, which do happen periodically
[07:50:51] <barque> the task state registers were there
[07:50:54] <barque> no one used them
[07:50:56] <ohsix> (be it a hw irq or app generated irqs
[07:51:02] <barque> no one used segmentation when it came either
[07:51:26] <barque> everyone just set 1 descriptor table with base at 0 and segment size at 0xFFFFFFFF
[07:51:45] <ohsix> thats one of the problems of continuous support of software from a time when they did "bad things"
[07:51:50] <kshishkov> ohsix: and Ukrainian Ministry of Education and Science once almost approved high school of Astrology, so what?
[07:51:56] <barque> Oh god
[07:51:58] <barque> did that happen?
[07:52:00] <barque> oh shit
[07:52:12] <barque> the Kansas state board once taught creationism
[07:52:19] <ohsix> kshishkov: that sucks, but is a redherring
[07:52:21] <barque> take that in ignorance you ruskies
[07:52:39] <ohsix> teaching evolution is still illegal there
[07:53:18] <barque> lol 1 year of complete idiocy in American elementary schools
[07:53:27] * barque loves his home, Canada
[07:53:36] * av500 blames Canada
[07:53:40] <barque> lol
[07:53:41] <ohsix> the law was from the reagan era iirc
[07:53:48] <barque> lol no it was 2005
[07:53:54] <barque> and they abandoned it a year later
[07:53:55] <KotH> barque: and the state of colorado once had a vote whether they want to define Pi as 4
[07:54:01] <barque> once they realized the retardation involved
[07:54:05] <barque> oh yeah the Pi bill
[07:54:07] <barque> that was hilarious
[07:54:13] <barque> that was old though
[07:54:23] <barque> what was funny was that the senators were all excited
[07:54:25] <barque> hahahahaha
[07:54:30] <kshishkov> barque: may I remind that Canada got best Ukrainians?
[07:54:31] <ohsix> did they also propose to change the shape of a circle?
[07:54:32] <barque> seriously, what kind of goof balls do you guys elect?
[07:55:00] <ohsix> i don't hink anyone from colorado or kansas is here
[07:55:17] <KotH> barque: http://en.wikipedia.org/wiki/Christoph_Blocher
[07:55:31] <KotH> oh wait.. he got thorwn out again
[07:55:37] <av500> http://www.snopes.com/religion/pi.asp
[07:55:46] <av500> alabama, 3, false
[07:56:07] <barque> tight immigration?
[07:56:10] <barque> what an idiot
[07:56:24] <barque> seriously, hatred of immigration is simply racism
[07:56:30] <barque> and anti-capitalism and statism
[07:56:37] <barque> and this guy called for free markets? my ass
[07:56:43] <barque> what a living paradox
[07:56:49] <ohsix> i can't help but notice; while you are a colorful character, theres a lot of not-ffmpeg-devel going on
[07:56:59] <wbs> ohsix: I was just about to say that
[07:57:10] <barque> oh yes this will end now :|
[07:57:33] <ohsix> i do a bit of it myself :]
[07:57:53] <barque> and btw that was the Indianna Pi bill
[07:57:53] <wbs> ohsix: you don't type 10 times more lines than everybody else combined, at least
[07:57:55] <barque> just for the record
[07:58:09] <ohsix> barque: i got family that flat out hates mexicans but loves the foodl cognitive dissonance is funny
[07:58:20] <ohsix> wbs: not here :O
[08:00:09] <av500> ok, Indiana in the end: http://scienticity.net/wiki/Legislating_the_Value_of_Pi
[08:00:23] <av500> oops
[08:01:20] <av500> back on topic, is there anybody that *understands* compute_pkt_fields()?
[08:01:36] <ohsix> theres some irony to making "evolution" illegal; even if its just in the schools :]
[08:01:48] <KotH> av500: of course there is ;)
[08:01:58] <av500> KotH: don't say it
[08:02:09] <kshishkov> av500: look into Wien direction, there lives a guy who is rumoured to understand all FFmpeg code
[08:03:42] <barque> LOL
[08:03:52] <barque> sounds like S.T.A.L.K.E.R hermit
[08:04:13] <barque> legend has it, he lives at Old Warehouses
[08:04:22] <av500> [09:29] * barque goes home
[08:04:24] <jai> there are legends about this person as well :)
[08:05:01] <kshishkov> av500: your point being?
[08:05:17] <KotH> kshishkov: i think these rumours are wrong... ;)
[08:06:20] <kshishkov> smite the infidel!
[08:21:24] <elenril> av500: gcc understands it!
[08:21:29] * elenril runs
[08:27:47] <barque> lol
[08:27:56] <barque> gcc understands a lot of things
[08:28:40] <kshishkov> elenril: you fail Philosophy, since an ability to manipulate is not enough for understanding
[08:29:13] <spaam> kshishkov: have anyone seen this guy? :)
[08:29:16] * elenril fails many things
[08:29:29] <elenril> spaam: KotH claims he did
[08:29:48] <elenril> but he's an UnreliableNarrator, so we can't be sure
[08:29:54] <spaam> elenril: but that is KotH ... ;S
[08:46:18] * barque starts to really appreciate -cropleft and -cropright with -o whatever[%d].fmt
[08:46:38] <kshishkov> too late
[08:46:44] <barque> I probably will work a good deal on avformat
[08:46:59] <barque> but I would honestly very much so appreciate rapid review and commit on behalf of the community
[08:47:10] <barque> so we can make good progress
[08:47:20] <av500> barque: send patches
[08:47:32] <barque> please oh god please send a confirmation email
[08:47:34] <superdump> that depends on the quality of your code and implementations
[08:47:38] <barque> for the love of god I've sent 2 emails
[08:48:02] <barque> superdump, I don't mind community critique at all, just a bit... faster
[08:48:06] <barque> :]
[08:48:31] <av500> barque: rejected - now that was fast, no?
[08:48:47] <barque> rejected? what?
[08:48:55] <av500> your 1st patch
[08:48:59] <kshishkov> av500: it's not proper review, try "it sucks, rejected"
[08:49:29] <av500> kshishkov: I am being fast and polite
[08:49:29] <barque> it was just 3 lines
[08:49:38] <barque> and umm no it doesn't suck
[08:49:46] <barque> unless BGRA means magical things on magical machines
[08:50:10] <kshishkov> it is
[08:50:12] <barque> but umm yeah I'm wondering about confirmation emails right now
[08:50:14] <lu_zero> good morning
[08:50:27] <kshishkov> buon giorno
[08:51:08] <barque> and what I mean by faster is not... let's all whine faster
[08:51:14] <barque> let's all do faster
[08:51:21] <barque> all together now :]
[08:51:42] <barque> unless we want to move like elephants
[08:53:15] * kshishkov finds moving like an elefant quite acceptable
[08:53:38] <lu_zero> kshishkov: I'll bring the peanuts next time =P
[08:54:17] <KotH> barque: what confirmation mail?
[08:54:22] <av500> ml signup
[08:54:26] <barque> yeah
[08:54:27] <av500> if there is any
[08:54:27] <barque> mailing list
[08:54:35] <barque> ffmpeg-devel
[08:58:22] <johnf> is it possible to determine why av_open_input_stream failed?
[09:01:40] <Dark_Shikari> siretart: fyi, from now on with releases and x264, just grab the latest stable branch.
[09:02:43] <twnqx> will you release versioned stable sourceballs?
[09:02:58] <Dark_Shikari> no
[09:03:03] <Dark_Shikari> http://git.videolan.org/?p=x264.git;a=summary
[09:03:08] <Dark_Shikari> stable is the second green tag on this screen.
[09:03:08] <Tjoppen> johnf: yes, sort of. I check for AVERROR_EIO and report 404, otherwise I report "failed to parse header"
[09:03:11] <Dark_Shikari> Always.
[09:03:12] <Dark_Shikari> =p
[09:03:13] <barque> is there any sort of auto-crop in avformat?
[09:03:23] <barque> rather I should ask this in the other channel
[09:03:28] <Dark_Shikari> barque: that would be in avfilter if it existed.
[09:06:18] <siretart> Dark_Shikari: cool! thanks!
[09:06:34] <Dark_Shikari> rolling stable branches are so cool
[09:06:39] <Dark_Shikari> it's like a "free" stable branch
[09:06:45] <Dark_Shikari> unlike a stable branch, you don't have to actually maintain it!
[09:07:20] <siretart> Dark_Shikari: will work on the x264 package just after I've finished ffmpeg 0.6. currently blocked by mru's OK or patch :-)
[09:10:18] <Dark_Shikari> great, 0.6 will be outdated by the time it's released =p
[09:11:01] <barque> lol this is what I mean
[09:11:28] <ohsix> git ftw
[09:11:42] <barque> I dunno , the more I use FOSS the more I find the urge to contribute
[09:11:52] <barque> and the more I feel quadratic fluid drag
[09:11:59] <siretart> Dark_Shikari: it is already. but I think that's okay.
[09:12:35] <siretart> code doesn't oxidize, after all
[09:12:52] <av500> maybe chrome code?
[09:13:03] <av500> at least it needs polishing
[09:13:23] <siretart> possibly
[09:14:52] <CIA-92> ffmpeg: bcoudurier * r23608 /trunk/ (libavformat/isom.c libavformat/movenc.c libavcodec/raw.c): Support gray16be and rgb48be in mov
[09:15:11] <kshishkov> siretart: but it rots. Try compiling very old code on new machine. Unless it's evergreen "Hello, world!" it has a good chance of failing
[09:15:53] <Dark_Shikari> well not just that
[09:16:00] <Dark_Shikari> if you have code that works perfectly well one day
[09:16:02] <Dark_Shikari> and wait 5 years
[09:16:07] <Dark_Shikari> and compile it on the same machine in the same environment
[09:16:08] <Dark_Shikari> it'll probably fail
[09:16:09] <siretart> kshishkov: yes, but that mostly isn't because of defects in old code, but because it's environment changes.
[09:16:29] <av500> kshishkov: also, "hello" might become unpolite in the future :)
[09:16:37] <siretart> kshishkov: I don't think neither 0.5 nor the upcoming 0.6 suffer from this problem at this point
[09:16:51] <siretart> s/don't//
[09:17:34] <siretart> I'm looking forward to 0.7 and its hopefully fully functional libavfilter!
[09:17:53] <kshishkov> it will take a year :P
[09:17:54] <spaam> siretart: next year? ;)
[09:18:04] <siretart> spaam: that would be at least my plan, yes
[09:18:04] <kshishkov> and longer for avfilter
[09:18:44] <siretart> if there is demand for a release this fall, we can discuss this as well, but I'd really like to focus on ffmepg 0.6 and mplayer rc4 for now.
[09:19:18] <siretart> afk lunch
[09:27:42] <barque> lol Dark_Shikari, did Steve Jobs quote you once?
[09:27:56] <barque> haha nice, I read that article of yours as well
[09:27:58] <barque> nice analysis
[09:28:01] <Dark_Shikari> a few weeks ago, during his normal trolling procedures.
[09:28:07] <barque> heh
[09:28:37] <barque> generally all the sheningans in the world of cosine transform/wavelet compression has completely pushed me to the uncompressed world
[09:29:17] <barque> and to make it all worse, they add lossless compression/decompression like deflate/inflate on top
[09:29:21] <barque> how dare they :P
[09:29:26] * Tjoppen gives barque some DPX
[09:29:44] <barque> X_X
[09:29:50] * barque dies of a heart attack
[09:30:45] <Tjoppen> I like how I only managed to mux about one second worth of it in avi since it crosses the 2 GiB limit after that
[09:35:23] <barque> Dark_Shikari, out of curiosity, what made you interested in motion jpeg or ... video encoding/decoding in general
[09:36:22] <Dark_Shikari> motion jpeg?  who the hell cares about mjpeg
[09:37:05] <barque> ok mpeg
[09:37:10] <barque> whatever, videos... :D
[09:42:20] <kshishkov> barque: that's how your patch would look if you tried it harder - http://ffmpeg.pastebin.com/S5HHMXuR
[09:43:06] <barque> where is the 0x08
[09:43:08] <barque> that is needed
[09:43:32] <barque> that's basically my patch minus 0x08 needed by the TGA spec
[09:43:50] <kshishkov> where's that spec?
[09:44:16] <barque> http://local.wasp.uwa.edu.au/~pbourke/dataformats/tga/   scroll down to unmapped RGB
[09:44:34] <barque> bits 3 - 0
[09:44:52] <barque> and the format specifier is completely wrong, TGA has BGRA, not RGBA
[09:46:01] <kshishkov> "This entire byte should be set to 0.  Don't ask me." hmm...
[09:46:06] <barque> "unmapped"
[09:46:10] <barque> you're at the wrong section
[09:46:15] <barque> that's color mapped
[09:47:25] <kshishkov> http://samples.mplayerhq.hu/image-samples/TGA/rgb32rle.tga <- looks like it works without that attribute
[09:47:32] <barque> *sigh*
[09:47:37] <kshishkov> also that's not the official spec
[09:47:48] <barque> it's not, but I can see it in actual work
[09:47:53] <barque> when I work with GIMP
[09:48:40] <barque> you need to specify that you're using 8 bits in attribute, and as it says BGRA... GIMP gives me this
[09:48:44] <barque> I'm using it in a game engine
[09:49:00] <barque> why do you think I needed ffmpeg in the first place? :D
[09:55:02] <KotH> barque: watching animes
[09:55:14] <Dark_Shikari> there is no s in animes
[09:55:36] <ohsix> anime no wa
[09:56:17] <KotH> Dark_Shikari: that's why it is at the end ;)
[10:00:37] <spaam> lu_zero: if you merge libsws into ffmpeg-git..  how do mplayer get updates from libsws ?
[10:00:53] <mru> not
[10:01:09] * KotH flames mru for breaking mplayer, again
[10:01:42] <Dark_Shikari> so anyone else going to be at e3 tomorrow?
[10:01:47] <kshishkov> MPlayer is a hack, it's destined to break
[10:01:59] <spaam> haha
[10:02:11] <Dark_Shikari> or am I the only representative of FOSS multimedia there
[10:02:37] <mru> Dark_Shikari: would you like to do a 16x16 SAD in 8 cycles?
[10:02:53] <Dark_Shikari> mru: that would be great, especially if it's synchronous.
[10:03:07] <mru> async of course
[10:03:23] <Dark_Shikari> you already told me about this
[10:03:28] <Dark_Shikari> it's another asynch functional unit with a thousand cycle latency
[10:03:46] <mru> so do ME for the entire frame in one call or whatever
[10:04:07] <Dark_Shikari> yes, that requires restructuring the encoder
[10:04:23] <barque> Dark_Shikari, you go get'em
[10:04:28] <barque> do you have a booth
[10:04:29] <barque> ?
[10:04:30] <Dark_Shikari> why can't they just give me sadblock outputreg, src1, src1stride, src2, src2stride
[10:04:37] <Dark_Shikari> it's only 5-operand!
[10:04:47] <mru> it's not part of the ARM cpu
[10:04:54] <Dark_Shikari> neither is neon
[10:04:55] <barque> if you're showing up next E3 I might see you
[10:04:58] <mru> neon is
[10:05:03] <barque> I'll bring you food too
[10:05:54] <Dark_Shikari> mru: also, depends how long a cycle is
[10:06:04] <Dark_Shikari> and I don't see why it has to be so high latency
[10:06:13] <Dark_Shikari> why can't I just call it while doing intra analysis for that MB
[10:06:14] <mru> I haven't told you any latencies
[10:06:16] <Dark_Shikari> and get back the results when it's done?
[10:06:20] <Dark_Shikari> then what is it?
[10:06:22] <Dark_Shikari> 10? 100? 1000?
[10:06:26] <mru> I don't know
[10:06:28] <Dark_Shikari> lol
[10:06:40] <ohsix> subpel is a tradeoff against that sort of thing
[10:06:42] <Dark_Shikari> barque: no booth, we have a private room.  we're going to be inking deals while everyone else parties.
[10:07:00] <ohsix> fwiw my dpi is set right, too
[10:10:16] <barque> so you have clients showing up for contracts ? or you're going to gather clients?
[10:10:54] <Dark_Shikari> clients showing up for meetings/demos
[10:11:39] <barque> very nice
[10:11:49] <barque> of all places, E3? or were they planning on being there?
[10:12:05] <Dark_Shikari> of course at e3.
[10:12:21] <Dark_Shikari> E3 has never been about marketing to consumers
[10:12:28] <barque> true
[10:12:39] <merbzt> Dark_Shikari: licensing stuff complete?
[10:12:41] <barque> but I've never seen any private side of it
[10:12:49] <Dark_Shikari> merbzt: mostly, we have a license ready but I need comments on it from customers
[10:12:57] <barque> usually public boasting about this or that or what-have-you
[10:13:01] <merbzt> ok
[10:13:27] <barque> so you're doing large scale multimedia distribution or somesuch
[10:13:54] <barque> good stuff since you're sooooo damn into it... I just never caught me eye that much
[10:13:57] <Dark_Shikari> merbzt: so if you're a potential customer, ping me =p
[10:14:03] <barque> I definitely can see a bajillion good uses
[10:14:38] <barque> but meh... I'm more into triangular rasterization and its potential intricacies than anything else :]
[10:15:00] <barque> must have something to do with our developmental stages ;D
[10:15:08] <CIA-92> ffmpeg: mru * r23609 /trunk/tests/ref/fate/vc1: FATE: update vc1 reference output
[10:15:51] <barque> SVO seems like a good contender but definitely not in the motion department
[10:18:23] <Tjoppen> depends on how you render it.. I had an idea where you transform the space as you traverse the skeleton, then you render the voxels belonging to the bone you're on using the current transformation
[10:18:37] <barque> hah nice try
[10:18:38] <barque> but no
[10:18:44] <Tjoppen> yeah, I suspect as much :)
[10:18:48] <barque> :D
[10:18:54] <barque> here's what you need to do
[10:19:05] <barque> feedback loop between rendering buffer and your processing code
[10:19:14] <barque> as LOD changes without user interaction
[10:19:18] <barque> thrashing of the harddrive
[10:19:36] <barque> completely useless, since with triangular rasterization you get per-vertex transformations and hardware interpolation
[10:19:41] <barque> much more efficient
[10:19:48] <barque> "sorry John"
[10:20:44] <barque> I guess John is better off blowing up rockets
[10:20:46] <Tjoppen> voxels for static geometry (free geomod \o/ ), triangular meshes for everything else
[10:20:48] <barque> Hahahahahahaha that was so funny
[10:20:56] <barque> omg I'm such a comedian
[10:21:08] <barque> man I don't laugh at my own jokes but
[10:21:11] <barque> god damn it
[10:21:16] * barque pats self on the back
[10:23:00] <barque> god damn guy sold id
[10:23:03] <barque> did you get the news?
[10:23:18] <barque> id is now a subsidiary of Zenimax (publishers of Fallout 3)
[10:23:34] <barque> can it get any more pathetic
[10:23:54] <ohsix> now?
[10:24:00] <barque> well a whiiiile back
[10:24:11] <Tjoppen> "news"
[10:24:15] <ohsix> i think them holding onto gamebryo for as long as they did was the pathetic part
[10:24:15] <Tjoppen> that was like three months ago
[10:24:28] <barque> Gamebryo was the most pathetic engine I've ever seen at work
[10:24:29] * kshishkov waits for sucky new Commander Keen as FPS
[10:24:38] <Tjoppen> speaking of games and e3: http://www.cracked.com/article_18608_the-day-gaming-industry-died-impressions-from-e3-2010.html
[10:25:00] <barque> oh my god lol
[10:25:01] <barque> so true
[10:25:08] <barque> I was watching the feed yesterday from youtube
[10:25:17] <barque> I could not decide whether to laugh or cry
[10:26:47] <barque> speaking of Gamebryo, I could've done better skeletal animating alone with Blender than what they did in Fallout 3
[10:27:03] <barque> they really should've given me a call
[10:27:12] <barque> me doing their models in my spare time, would've been better than what they did
[10:27:34] <ohsix> my favorite thing about gamebryo is how it handles dead/idle objects; when you wake them up they levitate
[10:27:35] <barque> and I couldn't critize it , in fear of being beaten up by 15 year old gamestop store workers
[10:27:57] <Tjoppen> ohsix: yeah, that bugged me a lot
[10:28:20] <ohsix> its pretty silly to see over and over if you're the skulking type :\
[10:28:22] <barque> you know why they levitate?
[10:28:38] <ohsix> yes
[10:28:39] <barque> I'll give you a shot at it
[10:28:42] <barque> why? :)
[10:29:04] <ohsix> it would be better if they lerped to as moved instead of
[10:29:17] <ohsix> "near enough to move"
[10:29:17] <barque> do you know *why* they levitate
[10:29:44] <ohsix> so they don't explode off the surface
[10:29:58] <ohsix> cuz gamebryo suxxx
[10:30:01] <barque> because if rigid body is on the floor and near triangle edge, since their collision detection is worse than Bullet's, you wouldn't get a neat slide on the surface
[10:30:16] <barque> and it'd look like my ass jittering on the floor
[10:30:24] <barque> against flat surface
[10:30:30] <barque> so they move it up to avoid that intersection
[10:30:34] <barque> so you can throw it away
[10:30:53] <barque> and lack of surface friction was even worse
[10:31:03] <barque> this is why I hate 3rd party multirigid body simulators
[10:31:13] <barque> they either do a bad job of friction, or don't even bother
[10:32:05] <ohsix> with gamebryo its already a losing proposition; they don't much have a budget for much with the way it runs
[10:32:18] <barque> heh good
[10:32:20] <barque> more room for me
[10:32:28] <barque> the more these dinasours die
[10:32:36] <barque> they better and more vibrant and diverse the market becomes
[10:42:14] <wbs> why is it that 85% of the lines written while I was away for lunch was written by one single person?
[10:42:26] <wbs> and has about how much to do with ffmpeg-development?
[10:45:27] <kshishkov> wbs: lowering channel SNR
[10:46:01] <wbs> kshishkov: yes, I guess it's good for improving people's noise filtering abilities ;P
[10:52:41] <johnf> I have some code at the moment whch does  av_guess_format("ffm", NULL, NULL) which fcreates ffm container with mpeg1video. I want to set it to be ffm with different video codec inside. Can someone point me at code on how to generate a AVOutputFormat with the right codecs?
[10:53:41] <av500> -> #ffmpeg_devel
[10:53:45] <av500> err
[10:53:47] <av500> -> #ffmpeg
[10:54:59] <johnf> opps orry
[12:13:30] <enkidu> it seems, that ffserver is not closing particular streams. I tuned tcp_timeout to one minute, and still I can see hanging connections.
[12:14:24] <enkidu> HTTP/1.1 WAIT_FEED 828k 0 420
[12:30:08] <CIA-92> ffmpeg: siretart * r23610 /branches/0.6/ (configure libavformat/utils.c):
[12:30:09] <CIA-92> ffmpeg: add symver hacks for gnu linkers
[12:30:09] <CIA-92> ffmpeg: this patch restores binary compatibility for the av_*_packet symbols that have
[12:30:09] <CIA-92> ffmpeg: been moved from libavformat to libavcodec. This patch works for gnu toolchains
[12:30:09] <CIA-92> ffmpeg: only; support for ARM RCVT will be handed in for a later point release as soon
[12:30:09] <CIA-92> ffmpeg: as the patch is ready and approved by the ARM maintainer(s).
[12:33:20] <mru> siretart: why'd you do that?
[12:33:31] <mru> I'm 10 minutes away from a good patch
[12:34:21] <siretart> mru: oh, I didn't know that. I had the impression that you worked on something else
[12:34:29] <mru> you could have asked
[12:34:35] <siretart> didn't I
[12:34:36] <siretart> ?
[12:34:47] <mru> and I said I'd do it after linuxtag
[12:34:50] <mru> now is after linuxtag
[12:35:03] <siretart> the next 3 weeks is after linuxtag as well ;-)
[12:35:14] <siretart> anyway, sorry about that then, I'll of course revert my patch then
[12:35:29] <av500> after linuxtag is also before linuxtag
[12:35:39] <siretart> I'd say just commit to trunk, and I'll handle the backport to 0.6, ok?
[12:35:48] <mru> sure
[12:35:59] <mru> branches are your jungle
[12:36:50] * KotH imagines siretart with a topee and a machete
[12:38:42] <siretart> KotH: :-)
[12:54:15] <enkidu> the sockaddr_in structure contains sin_port, which is peer or host port number?
[12:54:58] <wbs> depends on which sockaddr_in you're talking about
[12:56:00] <enkidu> wbs, the one in compute_status, where sin_addr contains peer IP
[12:56:47] <wbs> I'm not that familiar with the ffserver code, and I don't have time to check right now - trace where it comes from
[12:57:18] <enkidu> ok, ill check it myself then ;]
[12:57:20] <wbs> ah, well if the sockaddr_in contains the peer ip, then the port obviously also is the port on the peer
[13:03:07] <lu_zero> wbs: j0sh_'s patch for vlc is somewhere I'm missing or it's still in his hd and just there?
[13:04:06] <wbs> lu_zero: didn't he send it to -soc a few days ago?
[13:04:58] <wbs> lu_zero: http://lists.mplayerhq.hu/pipermail/ffmpeg-soc/2010-June/009265.html
[13:05:08] <lu_zero> thunderbird archived in the wrong place...
[13:05:11] <lu_zero> found
[13:07:58] <enkidu> ill commit two patches for status page
[13:08:13] <enkidu> s/commit/send to group/g
[13:15:06] <mru> siretart: new patch sent
[13:15:09] <mru> smaller than yours
[13:20:06] <siretart> on my laptop, configure sets SYMVER 1, SYMVER_GNU_ASM 1, SYMVER_ASM_LABEL 0. looks great, testbuild still running
[13:23:33] <enkidu> should I try webm codec?
[13:23:47] <mru> why would you want to do that?
[13:23:47] <siretart> mru: works like a charm! thanks!
[13:24:37] <enkidu> mru: because I would like to provide higher resolution streams?
[13:24:52] <mru> and h264 isn't good enough?
[13:25:25] <enkidu> too heavy + needs flash for web payer
[13:25:26] <elenril> everybody knows that webm is better
[13:25:39] <mru> google chrome supports h264
[13:26:05] <siretart> mru: please commit, I'll then backport and prepare the 0.6 tarballs :-)
[13:26:12] <janneg> enkidu: webm/vp8 also needs a flash player for safari/IE
[13:26:37] <elenril> anyone knows what happened to MAX_STREAMS removal?
[13:26:38] <enkidu> janneg: for safari and IE. most people here are using FX/Opera
[13:26:43] <janneg> and released chrome/firefox/opera versions
[13:27:10] <janneg> webm support is only in pre-releases
[13:27:42] <enkidu> yeah... thats why I wanted to wait with it after summer holidays
[13:27:44] <CIA-92> ffmpeg: mru * r23611 /trunk/ (libavutil/internal.h configure libavformat/utils.c):
[13:27:44] <CIA-92> ffmpeg: Add compatibility wrappers for functions moved from lavf to lavc
[13:27:44] <CIA-92> ffmpeg: When symbol versioning is enabled, moving symbols from one library to
[13:27:44] <CIA-92> ffmpeg: another breaks binary compatibility. This adds wrappers with the old
[13:27:44] <CIA-92> ffmpeg: version tag for the av_*packet functions recently moved to lavc.
[13:27:51] <siretart> \o/
[13:28:45] <wbs> chrome prereleases doesn't seem to support streamed webm too well though. firefox works so-so, requiring that you let it buffer "long enough" before pressing play. opera works quite well though
[13:29:38] <av500> wbs: it says "buffering..." at least?
[13:29:46] <wbs> av500: which one?
[13:30:01] <av500> ff
[13:30:05] <enkidu> anyways, there really is possibility to view peer port
[13:30:09] <CIA-92> ffmpeg: siretart * r23612 /branches/0.6/ (configure libavformat/utils.c): revert r23610, better patch is pending
[13:30:39] <wbs> av500: usually it just looks like it is stuck
[13:32:18] <CIA-92> ffmpeg: siretart * r23613 /branches/ (4 files in 4 dirs):
[13:32:18] <CIA-92> ffmpeg: Add compatibility wrappers for functions moved from lavf to lavc
[13:32:18] <CIA-92> ffmpeg: When symbol versioning is enabled, moving symbols from one library to
[13:32:18] <CIA-92> ffmpeg: another breaks binary compatibility. This adds wrappers with the old
[13:32:18] <CIA-92> ffmpeg: version tag for the av_*packet functions recently moved to lavc.
[13:32:19] <CIA-92> ffmpeg: backport r23611 by mru
[13:34:08] <Honoome> siretart: sorry if I haven't checked this yet, but do you need wrappers or they maintain the same ABI? if so you want aliases, not wrappers
[13:34:34] <mru> this has been discussed to death
[13:34:38] <mru> we need wrappers
[13:34:43] <Honoome> k
[13:34:47] <mru> you can't have an alias in a different shared object
[13:35:06] <mru> otherwise you'd be right
[13:35:42] <Honoome> mru: hmm I think glibc allows that, but that's as far as I can get, as I have no idea how much that's standard and/or if it works just out of sheer luck when it does :P
[13:35:50] <Honoome> [which considering it's glibc is _quite_ possible]
[13:36:01] <mru> no, it's not possible anywhere
[13:36:05] <siretart> michael claims this to be a bug in the runtime linker, and I agree that the behavior here is indeed undesireable. I'm going to ask around if this others agree that this is something that should be changed
[13:36:13] <siretart> but for now we have to live with this
[13:36:19] <mru> aliases work by defining multiple symbols with the same address
[13:36:30] <mru> that can obviously only work within one object
[13:37:02] <mru> michael has trouble distinguishing between actual bugs and how he'd like things to be
[13:37:32] <mru> and even if it is a bug, it won't be fixed on people's machines anytime soon
[13:38:14] <Honoome> unless drepper slips it in the next glibc release without documenting it anywhere beside for a post on the C++ WG mailing list…
[13:38:22] <Honoome> ah no sorry that was STB_GNU_UNIQUE
[13:38:42] <Honoome> [and no I wasn't kidding]
[13:38:44] <av500> Honoome: the dell behaves now?
[13:38:45] <mru> it would still take debian half a century to pick it up
[13:38:54] <siretart> hrmpf. why did diego just go offline ...
[13:39:02] <mru> which diego?
[13:39:08] <Honoome> mru: I guess not me ;)
[13:39:12] <siretart> dondiego
[13:39:30] <Honoome> av500: sorta … it suspends fine, still has trouble shutting down when AC power is connected… the smartcard reader works… touchpad doesn't
[13:40:03] <mru> I'd just return it for a refund
[13:40:03] <siretart> KotH: would you mind installing http://tauware.de/~siretart/ffmpeg-0.6/ to http://ffmpeg.org/releases, please?
[13:40:17] <kshishkov> mru: can you now determine the year when GNU stuff was at its best or it's still getting better?
[13:40:37] <Honoome> mru: I wanted to as well, but I waived that when I ordered it as a company rather than a private user :/
[13:40:37] <mru> depends on which gnu stuff
[13:40:51] <mru> companies have no returns?
[13:40:53] <mru> wtf?
[13:41:14] <av500> mru: yep
[13:41:15] <kshishkov> mru: core GNU tools including binutils and GCC
[13:41:16] <av500> same here
[13:41:27] <Honoome> not out of whim … since they never put in writing that Linux was supported on that laptop but just told me on the phone, I can't even blame them for selling me something out of specs
[13:42:28] <mru> sell it on ebay then
[13:42:38] <av500> or test if it blends
[13:42:48] <enkidu> ok, there is "supercomplicated" patch: http://pastebin.ca/1883594
[13:42:55] <mru> binutils are probably at peak now
[13:43:12] <mru> the whole gold thing is probably the first step of the downfall
[13:43:16] <Honoome> mru: with or without gold? :P
[13:43:18] <Honoome> ah there you go :)
[13:43:33] <Honoome> beside the touchpad the laptop is fine though… looks quite solid and it's bloody fast :)
[13:43:37] <mru> gcc is hard to say
[13:43:49] <mru> 4.3 is in many ways better than 4.2 but also has some serious regressions
[13:43:56] <mru> 4.4 is mostly worse than 4.3
[13:44:10] <mru> 4.5 fixes much of 4.4 and is often better than 4.3
[13:44:16] <mru> but also quite a few regressions
[13:45:07] <mru> with the c++ rewrite looming, I guess it's fairly safe to say we've seen the best it'll be for quite some time
[13:46:17] <BBB> ffcc is looming
[13:46:18] <BBB> ?
[13:46:40] <enkidu> mru: does my patch conform ffmpeg standard?
[13:46:57] <mru> if you have to ask the answer is probably no
[13:47:11] <kshishkov> what about gnu ABI and glibc?
[13:47:59] <KotH> siretart: done
[13:49:16] * elenril wonders when will llvm be able to compile ffmpeg
[13:50:39] <janneg> elenril: somewhen in the past. see http://fate.multimedia.cx/index.php?build_record=241075
[13:51:05] <elenril> that's x86_64
[13:51:12] <mru> fuck
[13:51:13] <elenril> on 32bit it fails
[13:51:17] <mru> we broke windows builds
[13:51:29] <elenril> woot \o/
[13:52:26] <mru> Cannot export av_destruct_packet at LIBAVFORMAT_52: symbol not found
[13:54:01] <lu_zero> elenril: on x86_64 it compiles it fine
[13:54:38] <mru> ramiro: ping
[13:54:41] <elenril> lu_zero: read what i said =p
[13:54:53] <Honoome> otoh clang fails to build _anything_ for me with -ggdb
[13:57:11] <lu_zero> I think it fails to _link_
[13:57:29] <lu_zero> probably because you weren't using gold (obviously)
[13:57:31] <Honoome> I said build, not compile
[13:57:38] <Honoome> why? clang needs gold now? =_=
[13:57:58] <lu_zero> Honoome: I'm sure it doesn't but I would see a theme
[13:58:06] <BBB> mru: leading "_" missing?
[13:59:56] <mru> no idea
[14:00:30] <Honoome> hrm, I'm tired to fight with my mind to devise user acls in rails… =_=
[14:00:49] <elenril> lu_zero: no, it fails to compile
[14:00:57] <elenril> not enough registers or something
[14:03:53] <lu_zero> elenril: on x86 I'd expect that, even if we complain that much about gcc being dumb
[14:04:11] <lu_zero> mapping regs on that arch is not so trivial
[14:04:34] <lu_zero> still the issue with -ggdb is an actual bug
[14:05:19] <lu_zero> if really ends up that gold could understand the resulting elf that means that they just tried with it
[14:06:20] <Honoome> lu_zero: actually, it's not during link phase the problem
[14:06:22] <Honoome> it's gas that bails out
[14:08:20] <mru> siretart: will you hold off the release until we've fixed win32 builds?
[14:10:34] <siretart> mru: did we have win32 for 0.5?
[14:10:43] <siretart> win32 builds, even
[14:10:54] <av500> 0.6 is a good time to deprecate win32...
[14:10:58] <mru> no, but it would be nice if the code _can_ be built
[14:11:21] <siretart> mru: I'd prefer to hand that in with 0.6.1
[14:11:22] <janneg> siretart: win32 was building before the symbol versioning change
[14:11:36] <av500> BBB: btw, I found a new licence violator, its an android app but i has no homepage
[14:12:02] <siretart> janneg: in the sense that the symbol versioning change broke win32 builds, or is that coincidence?
[14:12:16] <mru> my commit broke win32 builds
[14:12:22] <janneg> yes, that commit broke it
[14:12:35] <mru> I don't have a windows machine so can't test anything there
[14:13:24] <siretart> janneg: can you fix the patch?
[14:13:31] <Honoome> grrrrrrrrr we start getting -flto bugs in gentoo
[14:13:45] <mru> does anyone here have build tools for windows?
[14:14:06] <janneg> siretart: I don't have a windows build machine
[14:14:36] <mru> siretart: please wait a few hours with the release
[14:14:50] <mru> it wouldn't look good if the release doesn't even build on win32
[14:14:58] <siretart> mru: okay, can you then remove the 0.6 files from the release directory?
[14:15:40] <mru> Vitor1001: do you have a windows machine?
[14:15:45] <mru> or am I imagining things?
[14:16:03] <Vitor1001> mru: No, what would I do with such a thing?
[14:16:05] <Vitor1001> ;)
[14:16:09] <mru> beats me
[14:16:22] <mru> then it's someone else who has one
[14:16:36] <elenril> can't you just install mingw32 and crosscompile?
[14:16:50] <mru> don't want to
[14:16:57] <mru> and I wouldn't even know where to begin
[14:17:14] <av500> mkdir mingw?
[14:17:26] <Honoome> crossdev mingw32
[14:17:42] <mru> and I still can't test it
[14:17:44] <siretart> KotH: can you please remove the 0.6 files from the release directory, it seems that win32 is broken :-(
[14:17:47] <Honoome> mru: emerge wine :P
[14:17:54] <mru> wine is not windows
[14:18:19] <Honoome> lu_zero: should I dedicate myself to kernel hacking and implement FIONREAD on sctp? :P
[14:19:17] <lu_zero> mru: as implementation for testing it works well
[14:19:36] <lu_zero> e.g. testing world of warcraft...
[14:19:49] <mru> I wouldn't use it to test linker quirks
[14:20:26] <BBB> av500: send it to me or to the issue tracker
[14:20:28] <lu_zero> Honoome: right now I'd need another pair of eyes to understand what's wrong with vlc or feng or both
[14:20:29] <BBB> av500: I'll get on it
[14:20:38] <BBB> av500: I need a link to the app store though
[14:20:40] <BBB> if possible
[14:20:56] <Honoome> lu_zero: you know that the underlying rtp part is not something I'm very fond of… can you bisect to find the problem at least?
[14:22:39] <lu_zero> Honoome: I'm tracking it using vlc+ffrtsp
[14:22:41] <siretart> I've now installed mingw on my laptop. what else beside --cross-prefix=i586-mingw32msvc do I need to pass to configure?
[14:23:06] <siretart> configure tells me to set target arch and OS, but I'm not sure what and how to do so
[14:24:14] <mru> --arch=x86 --target-os=mingw32 apparently
[14:24:27] <mru> but we really need a real windows machine for this
[14:24:43] * elenril has a virtual windows machine
[14:24:45] <av500> BBB: it is not in app store
[14:24:52] <BBB> android app store?
[14:25:12] <CIA-92> ffmpeg: rbultje * r23614 /trunk/libavformat/mmst.c:
[14:25:13] <CIA-92> ffmpeg: Fix missing logging context in a series of dprintf()s. Partially based on
[14:25:13] <CIA-92> ffmpeg: patch by Zhentan Feng <spyfeng gmail com>.
[14:25:17] <av500> BBB: http://www.google.com/search?q=rockplayerbase
[14:25:41] <av500> the apk is here: http://www.mediafire.com/?u3lwq2z5mzm
[14:25:57] <BBB> please send it by email
[14:25:59] <BBB> otherwise I forge
[14:26:37] <siretart> mru: true. I'm getting "error: no previous prototype for 'getc'" when trying to build
[14:27:49] <kshishkov> BBB: so mail keeps you from forging?
[14:27:56] <av500> BBB: email submitted to SMTP server...
[14:32:59] <BBB> forget :)
[14:33:03] <BBB> anyway
[14:33:18] <siretart> elenril: can you reproduce the build problem?
[14:33:52] * siretart has started up his virtualbox with windows xp and tries to figure out how to install msys/wingw/subversion
[14:34:13] <mchinen> Hi, anyway to see if a given AVCodexContext has been opened? couldn't find an obvious flag or variable with this info.
[14:34:23] <elenril> siretart: installing git atm
[14:34:46] <spaam> siretart: http://tortoisesvn.tigris.org/  nice svn client for windows.
[14:35:38] <BBB> mchinen: usually priv_data has been allocated, but there isn't a one-off way
[14:35:52] <BBB> mchinen: what's the exact use case?
[14:35:58] <Honoome> lu_zero: send me the url in query please, different computer here
[14:36:32] <mchinen> BBB: i'm creating a copy of the avformatcontext so the build_index call can be used in a seperate thread while decoding
[14:36:49] <mchinen> using the avcodec_copy_context call gets me most of the way
[14:37:06] <mchinen> but i only want to open the ones that have been in the original
[14:37:18] <lu_zero> rtsp://backstage.lscube.org/live/live_2.sd
[14:37:38] <BBB> mchinen: is this inside av*? or is this in an application?
[14:37:44] <BBB> mchinen: creating a copy is a really bad idea
[14:37:45] <mchinen> inside av
[14:37:59] <BBB> mchinen: better make it threadsafe, or require the app to take care of this, no?
[14:38:06] <BBB> maybe the app doesn't want an extra thread
[14:38:28] <mchinen> yeah, its a flag
[14:38:56] <mchinen> i looked at making it threadsafe
[14:40:01] <mchinen> but its tricky to lay down the critical sections correctly because the parts that read a packet and parse it are fairly spread out, and the user can dictate when these happen
[14:40:25] <mchinen> so its more than just protecting the ByteIOContext with mutex
[14:41:00] <mchinen> copies are a bad idea hm
[14:41:55] <BBB> well, imagine that the input is not a file, but a stream with a unique session key
[14:41:59] <BBB> you can't open it twice
[14:42:33] <BBB> I think it makes sense that while you load the index, it blocks. that is exactly why it is a separate function call and isn't done automatically, right?
[14:42:48] <BBB> we still want it to be done as-fast-as-possible, but I think it's totally OK if it blocks, and you wouldn't expect it to play.
[14:43:02] <BBB> especially since the primary use case here is things like editing
[14:43:18] <BBB> if it takes a second or two to load, that's ok
[14:43:35] <mchinen> it could take a minute for large files
[14:43:57] <BBB> the thread doesnt' speed that up, does it?
[14:44:26] <BBB> I mean, you can't start your "edited fragment, starting at 00:06:26.04" until you've gotten the index
[14:44:37] <BBB> (which you safe anyway)
[14:44:42] <BBB> save
[14:45:02] <mchinen> it lets playback and work at the beginnning of the file to happen immediately if the client takes care of it tho
[14:45:29] <BBB> ramiro: isn't MANGLE() supposed to do the leading underscore?
[14:47:19] <mchinen> do we want to allow build indexes for the streaming cases as well?
[14:53:12] <mchinen> BBB: anyway thx.  I've got it mostly done so I just want to try it out, but I won't focus on it for long if it causes troubles.
[14:56:33] <Tjoppen> BBB: http-poke :)
[15:03:30] <BBB> mchinen: I think it's something that won't be easiest
[15:03:42] <BBB> mchinen: so maybe focusing on something else, or better yet, discuss it on ML, is a better idea
[15:03:45] <BBB> (as a start)
[15:03:48] <BBB> Tjoppen: thanks :-p
[15:04:45] <mchinen> BBB: ok, will do thx
[15:06:48] <Tjoppen> re-tested the patch a few days ago. it's still good, except the refactoring conflicted with the new code
[15:07:58] <BBB> Yuvi: where is the bilin filter in the spec? am I correct that it's missing?
[15:11:14] <Honoome> lu_zero decided to use the debian randomizer …
[15:11:44] <lu_zero> debian randomizer was 4
[15:11:47] <lu_zero> mine is 0
[15:11:53] <enkidu> int random() { return 4; //as resultof dice roll }
[15:11:55] <Honoome> ah sorry
[15:12:48] <enkidu> anyways, debian devs are paranoic
[15:13:14] <enkidu> they patched openssl to initialize some variables with zero. these variables were passed as random seed...
[15:15:26] <av500> http://www.gapingvoidgallery.com/product_info.php?products_id=1593
[15:16:39] * elenril wonders wtf is with configure running for over 15 minutes now
[15:16:59] <janneg> windows?
[15:17:11] <av500> ... trying all possible integers.... pass
[15:18:36] <elenril> damn, forgot to install yasm
[15:19:11] <elenril> ...and swscale
[15:19:26] <elenril> srsly, we should switch to git
[15:21:38] <kshishkov> not until GSoC ends
[15:25:20] <CIA-92> ffmpeg: mru * r23615 /trunk/libavutil/internal.h: Fix symbol version compat wrappers on systems with export prefixes
[15:26:41] <elenril> aren't the students using git anyway?
[15:27:20] <enkidu> they are
[15:27:58] <enkidu> for me (student too, not taking part in GSoC) git is even easier
[15:28:43] <mru> gsoc students don't commit to the main repo so for them it's irrelevant
[15:29:21] <siretart> mru: did you manage to test r23615 on a windows machine?
[15:29:27] <mru> no
[15:29:33] <mru> ramiro says it's correct
[15:29:45] <mru> let's see what fate says
[15:32:12] * siretart crosses fingers
[15:39:07] <mru> oh ffs, now icc 10.1 is failing
[15:39:52] <spaam> noo ;(
[15:42:03] <janneg> mru: also r23611
[15:42:13] <mru> yes, of course
[15:45:47] <siretart> well, mingw seems fixed now: http://fate.multimedia.cx/index.php?build_record=241228
[15:45:53] <mru> yeah
[15:46:15] <mru> all the BSDs being down makes me uneasy
[15:46:29] <av500> really=
[15:46:31] <av500> really?
[15:46:57] <mru> a ton of fate machines are not reporting
[15:46:57] <siretart> hm. I need to run home now, I see to come back online in a few hours
[15:47:13] <lu_zero> uhmmm
[15:47:44] <lu_zero> which is the right way to pass options to muxers?
[15:49:03] * lu_zero wanted to let users set base_timestamp
[16:09:47] <lu_zero> AVOutputFormat uses somehow avoptions?
[16:10:36] * elenril never understood how avoptions work
[16:11:23] <mru> by magic
[16:12:13] <elenril> <insert Umineko reference here>
[16:12:28] <Dark_Shikari> Small bombs.
[16:12:57] <mru> anyone have icc 10.1 around?
[16:14:36] * elenril wonders if Dark_Shikari has a highlight on umineko
[16:14:38] <BBB> lu_zero: no, it doesn't, but it's built-in to the loggin context AVClass so we could use it in the future
[16:15:47] * mru doesn't even know what umineko means
[16:15:53] <mru> "neko" I've heard of...
[16:16:05] <Dark_Shikari> mru: "seagull"
[16:16:06] <elenril> an awesome novel you should read ASAP
[16:16:13] <twnqx> seacat!"
[16:16:15] <twnqx> :P
[16:16:24] <Dark_Shikari> seagull in japanese is literally "sea cat"
[16:16:29] <mru> hmm
[16:16:31] <twnqx> yeah i know
[16:16:32] <Dark_Shikari> For some weird reason.
[16:16:36] <Dark_Shikari> I'm not quite sure why.
[16:16:39] <twnqx> i inflicted the pain of watching the junk on myself.
[16:16:42] <av500> fish eating see snake anybody?
[16:16:47] <Dark_Shikari> twnqx: oh dear.  you watched the animu.
[16:17:13] <twnqx> i rated it 1.5 on anidb.
[16:17:36] <Dark_Shikari> twnqx: one word
[16:17:38] <Dark_Shikari> DEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEN
[16:17:46] <twnqx> :X
[16:17:54] <elenril> Dark_Shikari: because the sounds they make sound like meowing?
[16:18:04] <Dark_Shikari> "cawww caww"?
[16:18:30] <Dark_Shikari> twnqx: read the original.  adaptations suck, anime adaptations doubly so, deen adaptations five billion times so
[16:18:42] <elenril> btw i think only one kind of seagull is called that
[16:18:44] <twnqx> novel or manga?
[16:18:47] <Dark_Shikari> novel.
[16:18:56] <Dark_Shikari> manga is apparently actually decent, but I haven't read it.
[16:18:57] <twnqx> mh
[16:18:57] <lu_zero> BBB: I need a way to set the  base_timestamp
[16:19:00] <mru> there is only one kind of seagull
[16:19:06] <mru> other gulls are called other things
[16:19:20] <lu_zero> since having it random is a bit annoying for my purposes ^^;
[16:19:26] <elenril> The word umineko is the name of a kind of seagull known as a Black-tailed Gull.
[16:19:38] <mru> so then it's not the seagull
[16:19:45] <elenril> whatever
[16:19:47] <Dark_Shikari> mru: not quite true
[16:19:48] <Dark_Shikari> http://en.wikipedia.org/wiki/Seagull
[16:19:49] <lu_zero> <@Dark_Shikari> twnqx: read the original. <- so the game
[16:19:50] <elenril> чайка
[16:19:51] <Dark_Shikari> It's a family, not a species.
[16:20:09] <Dark_Shikari> lu_zero: It has no choices, no buttons, and is nothing but text, sound, and some images.  It's a book.
[16:20:21] <lu_zero> agreed
[16:20:38] <av500> last book I read had no sound
[16:20:43] <twnqx> oh, a visual novel.
[16:20:49] <Dark_Shikari> also, mru, this is one of my favorite images on wikipedia ever
[16:20:50] <Dark_Shikari> http://upload.wikimedia.org/wikipedia/commons/5/5f/Birdsniper.jpg
[16:20:55] <av500> a 500GB ipad app?
[16:20:56] <lu_zero> and is even written in a not so simple language
[16:21:07] <mru> Dark_Shikari: good one
[16:21:16] <Dark_Shikari> umineko 7: when the seagulls steal food.
[16:21:54] <janneg> the "wmv to avi converted" we saw friday on the ultra low res screen is at least no obvious license violator
[16:22:54] <av500> janneg: ?
[16:23:14] <Dark_Shikari> twnqx: get game, get translation from http://www.witch-hunt.com/ , read.
[16:24:18] <janneg> av500: diego insisted to view the world cup match friday evening
[16:24:39] <mru> at a place with the worst waiters ever
[16:26:11] <av500> you waited for them?
[16:26:50] <lu_zero> nah
[16:27:00] <lu_zero> they were at least trying to be sociable
[16:27:02] <janneg> they had a 3m 256x144 led display, after they stopped the tv playing app it had "wmv to avi converter" on the desktop
[16:27:27] <av500> thats a lof of leds...
[16:27:36] <mru> not enough
[16:29:32] <lu_zero> AVFormatParameters ...
[16:30:13] <janneg> and it had already a couple dead ones
[16:30:15] * elenril heard something about AVFormatParameters being deprecated a few years ago
[16:30:55] <lu_zero> AVOption might be better
[16:31:08] * lu_zero ponders about which options to use
[16:36:56] <Dark_Shikari> fffffffffffffffffffffff why does anyone use cvs
[16:37:16] <Dark_Shikari> whenever I see an email even mentioning an ongoing project using it I cringe
[16:41:04] <BBB> because it exists
[16:42:35] <elenril> s/cvs/svn/
[16:42:46] * mru cringes
[16:42:57] <BBB> let's switch to bzr
[16:42:58] <elenril> btw aren't you at e3?
[16:43:00] * BBB awaits kick
[16:43:07] <elenril> BBB: no, mercurial!
[16:43:09] <lu_zero> BBB: you didn't say hg
[16:43:10] <av500> BBB: what is wrong with sharing files on an ftp server?
[16:43:11] <BBB> \o/
[16:43:11] * mru kicks BBB
[16:43:11] <lu_zero> there
[16:43:19] * lu_zero kicks elenril 
[16:43:28] * janneg votes for monotome
[16:43:44] <lu_zero> janneg: there is a limit to masochism
[16:45:03] <janneg> lu_zero: I'm sure yesterdays release fixes all issues ;)
[16:46:47] <mru> monotone, aren't they they guys who change file format more often than they change their socks?
[16:46:55] <lu_zero> mru: no
[16:47:11] <mru> when did they stop doing that?
[16:47:45] <janneg> argh, C++ and a centralized db in ~/.mtm
[16:48:03] <mru> that's only the start of it
[16:48:41] <Honoome> mru: you confused mtm with bzr up there
[16:49:02] <lu_zero> for the 1st of april
[16:49:15] <mru> maybe bzr do it too
[16:49:24] <mru> I don't think anyone has a monopoly on being stupid
[16:49:29] <lu_zero> ffmpeg migrates to monotone as cms and to C++0x as implementation language
[16:49:34] <Honoome> mru: no but canonical comes close *cough*
[16:49:48] <lu_zero> well
[16:49:49] <mru> lu_zero: ffads would make a better joke
[16:49:56] <lu_zero> ffads?
[16:50:07] <mru> ads inserted in decoded video
[16:50:13] <lu_zero> oh
[16:50:20] <lu_zero> that would be interesting as well
[16:50:20] <janneg> our linuxtag business idea
[16:50:37] <mru> last year's was enterprise edition
[16:50:40] <mru> ffee
[16:50:49] <lu_zero> uhm
[16:50:59] <lu_zero> that wouldn't be that stupid to get founds
[16:51:13] <lu_zero> ffee <- last year release with additional dust
[16:51:14] <mru> it would be a great way to lose all users
[16:51:25] <lu_zero> well, yes
[16:51:52] * Dark_Shikari is off to E3
[16:52:06] <av500> kinetic monkey dancing?
[16:53:15] <janneg> steve ballmer was just testing early prototypes?
[16:53:28] <av500> they used him to calibrate
[16:53:38] <kierank> GET ON YOUR FEET
[17:09:56] <enkidu> subliminal messages? ;)
[17:17:03] <enkidu> I have just compared theora in mplayer / webm in firefox. theora still wins competition in smoothness of playback
[17:17:54] <janneg> I would blame firefox
[17:18:16] <janneg> have you tried the webm file in mplayer?
[17:21:11] <enkidu> not yet, but I would be glad if there is sample video compressed vith theora 1.1 and vp8
[17:54:23] <siretart> do we have an idea why icc 10 doesn't like ffmpeg trunk yet?
[17:55:40] <enkidu> icc doesnt like many thingies
[17:56:40] <mru> yes, I know why, and a fix is imminent
[18:02:08] <enkidu> is anyone working on ogg mux for theora live stream?
[18:03:04] * mru hopes not
[18:03:15] <enkidu> why "hopes not"?
[18:04:10] <mru> because it's shit
[18:05:04] <enkidu> nah, its not
[18:05:29] <mru> what's your reference? bink?
[18:05:35] <siretart> mru: awesome!
[18:05:57] <enkidu> mru: ive referenced movies made by OpenMovie
[18:06:07] <mru> sounds scary
[18:06:39] <lu_zero> enkidu: I'd rather have webm pseudostream
[18:06:39] <enkidu> x264 and theora1.1 seems to be competitive in medium-bitrate class
[18:06:41] <mru> siretart: pushed
[18:06:51] <mru> enkidu: ROTFL
[18:06:54] <lu_zero> enkidu: that would surprise me a lot
[18:06:55] <mru> Dark_Shikari: ping
[18:07:05] <lu_zero> Dark_Shikari: food for you!
[18:07:09] <mru> x264 totally crushes theora in any setting
[18:07:21] <CIA-92> ffmpeg: mru * r23616 /trunk/configure: (log message trimmed)
[18:07:21] <CIA-92> ffmpeg: configure: fix symver detection with icc 10
[18:07:22] <CIA-92> ffmpeg: Apparently the compiler sometimes bypasses the normal assembler, in
[18:07:22] <CIA-92> ffmpeg: which case the asm label variant works. If the file contains any
[18:07:22] <CIA-92> ffmpeg: inline asm, this bypassing is not done, and the assembler chokes on
[18:07:22] <CIA-92> ffmpeg: invalid syntax. Inserting a dummy asm() statement in the test when
[18:07:23] <CIA-92> ffmpeg: the compiler supports it makes sure no shortcuts are taken and the
[18:07:35] <siretart> mru: \o/ - let's see how fate likes this revision
[18:07:36] <enkidu> mostly in "using end user cpu power"
[18:07:38] <elenril> \o/
[18:08:03] <lu_zero> enkidu: not see how that changes
[18:08:16] <lu_zero> enkidu: did you also try mpeg2 while comparing?
[18:08:33] <siretart> mru: the commit message sounds like *facepalm* :/
[18:08:34] <mru> and mpeg4
[18:08:54] <mru> siretart: did you expect otherwise from icc?
[18:09:01] <mru> old version at that
[18:09:02] <lu_zero> and obviously vp8
[18:10:28] <enkidu> lu_zero: if I had vp8 example encoded from source files, then yes. No reason in comparing vp8 encoded from x264 with x264
[18:11:04] <mru> well, go get some uncompressed source files
[18:13:13] <lu_zero> enkidu: we have some bbb
[18:13:22] <lu_zero> janneg: the raw are still there?
[18:13:44] <mru> 1080p sources are available officially
[18:14:47] <janneg> lu_zero: not all source pngs are online
[18:16:44] <kierank> oh great yet another x264 theora comparison
[18:23:14] <BBB> isn't theora outdated now that vp8 is public?
[18:23:23] <lu_zero> depends
[18:23:27] <BBB> nobody cares about theora anymore, especially not its former fanboys
[18:23:28] <lu_zero> there are fans
[18:23:35] <BBB> they're now fanboy'ing around vp8/webm
[18:23:35] <lu_zero> BBB: some remain
[18:23:40] <lu_zero> nah
[18:23:43] <BBB> you've got to be kidding me? can I see? :)
[18:23:59] <kierank> theora is going to become the beos of video formats
[18:24:02] <BBB> I want to stick a broom up their back
[18:24:07] <lu_zero> kierank: NO
[18:24:25] <kierank> ?
[18:24:30] <lu_zero> beos was a nice idea with an incomplete implementation
[18:24:37] <mru> I guess he means it'll have one lingering user
[18:24:42] <kierank> yes i meant that
[18:24:52] <kierank> but that user will be very vocal
[18:25:22] <lu_zero> theora is a not so bright idea with an implementation that gets polished continuously
[18:44:46] <BBB> lu_zero: you're describing every second failure in silicon valley, and a few successes
[19:12:50] <siretart> mru: icc 10 seems work with r23616: http://fate.multimedia.cx/index.php?build_record=241343 :-)
[19:13:02] <mru> yeah, saw that
[19:13:16] <mru> I tested it myself before committing
[19:13:30] <mru> couldn't have figured out the reason otherwise
[19:18:25] <siretart> :-)
[19:19:32] <lu_zero> BBB: sad but possibly true
[19:19:50] <lu_zero> anybody willing to teach me how to add AVOption?
[19:22:43] <mru> ask options.c
[19:24:24] * lu_zero asked opt.c
[19:45:00] <CIA-92> ffmpeg: siretart * r23617 /branches/ (0.6/libavutil/internal.h 0.6):
[19:45:00] <CIA-92> ffmpeg: Fix symbol version compat wrappers on systems with export prefixes
[19:45:00] <CIA-92> ffmpeg: backport r23615 by mru
[19:45:00] <CIA-92> ffmpeg: siretart * r23618 /branches/ (0.6 0.6/configure): (log message trimmed)
[19:45:00] <CIA-92> ffmpeg: configure: fix symver detection with icc 10
[19:45:01] <CIA-92> ffmpeg: Apparently the compiler sometimes bypasses the normal assembler, in
[19:45:02] <CIA-92> ffmpeg: which case the asm label variant works. If the file contains any
[19:45:02] <CIA-92> ffmpeg: inline asm, this bypassing is not done, and the assembler chokes on
[19:45:03] <CIA-92> ffmpeg: invalid syntax. Inserting a dummy asm() statement in the test when
[19:45:03] <CIA-92> ffmpeg: the compiler supports it makes sure no shortcuts are taken and the
[19:54:05] <CIA-92> ffmpeg: aurel * r23619 /trunk/libavformat/matroskadec.c:
[19:54:06] <CIA-92> ffmpeg: matroskadec : propagate AVERROR to the caller
[19:54:06] <CIA-92> ffmpeg: based on a patch by Jai Menon
[19:54:15] <elenril> release now?
[19:55:28] <siretart> on it
[20:01:37] <siretart> elenril: http://wiki.tauware.de/~siretart/ffmpeg-0.6/
[20:02:11] <siretart> anyone around who could move the files in place?
[20:02:20] <elenril> \o/
[20:06:52] <_av500_> lift off?
[20:07:57] <siretart> still needs publishing at http://ffmpeg.org/releases, but I don't have access to the machine
[20:08:31] <mru> can you make them accessible in some sane way?
[20:09:23] <mru> guess wget -r does the job
[20:10:20] <siretart> what would be sane for you? lftp's mirror command should do even via http
[20:10:43] <mru> nvm, I got them
[20:11:59] <mru> all in place
[20:13:55] <siretart> \o/ thanks a lot!
[20:16:25] <wbs> ramiro: how do you set up the mingw build to avoid all the "error: no previous prototype for 'getc'" errors?
[20:16:41] <mru> maybe he patches the system headers
[20:17:23] <wbs> would be nice if such changes were pushed upstream, too... mingw32ce requires a bit of hacking of the system headers, but up until that problem, normal mingw has worked fine
[20:18:36] <mru> mingw seems mostly dead
[20:20:05] <wbs> that's a shame, it's very useful if you have to deal with that platform
[20:20:31] <mchinen> is anyone working on getting flac seek to work?
[20:20:45] <mru> wbs: cygwin is better
[20:20:49] <mru> linux better still
[20:20:53] <lu_zero> mchinen: hi
[20:21:01] <mchinen> lu_zero: hi
[20:21:09] <lu_zero> why you are putting patches in svn btw?
[20:21:12] * Honoome reads flac seek and runs away screaming
[20:21:31] <lu_zero> Honoome: .... he meant flac in ogg seek obviously
[20:21:42] <mchinen> i thought this was the correct way to do my soc work
[20:21:44] <wbs> better for some purposes, yes, not if building native windows apps is the target, though
[20:22:01] <mru> why would anybody want to do that?
[20:22:09] * elenril throws http://tvtropes.org/pmwiki/pmwiki.php/Main/GoMadFromTheRevelation at Honoome 
[20:22:17] <Honoome> lu_zero: ouch isn't that even worse?
[20:22:22] <Honoome> elenril: hmm not sure if that's the correct trope
[20:22:39] <Honoome> for me flac seeking is a continuous bane, one of the few things I never cared enough to implement in xine and people asked me to
[20:23:02] <mru> raw flac is one of the most awful file formats ever made
[20:23:06] <wbs> mru: well, that's the big question, yes ;P
[20:23:08] <mru> it's right up there with ogg
[20:23:14] <elenril> http://tvtropes.org/pmwiki/pmwiki.php/Main/ThisIsWrongOnSoManyLevels then?
[20:23:14] <_av500_> raw flac seeks fine here
[20:23:14] <mchinen> is ffmpeg using libflac?  I see some references to it on the web but still haven't seen the headers anywhere in flacdec.c etc
[20:23:18] <Kovensky> * +elenril throws http://tvtropes.org/pmwiki/pmwiki.php/Main/GoMadFromTheRevelation at Honoome  <-- should've been BigNo
[20:23:48] <mchinen> _av500_ hm.  in ffplay too?
[20:23:50] <Honoome> elenril: yeah that sounds good
[20:23:56] <_av500_> mchinen: no
[20:24:03] <_av500_> in my own player
[20:24:13] <mru> real hackers make their own players
[20:24:45] * mchinen fake hacker
[20:25:28] * mru is playing flac with his own player right now
[20:26:03] <mchinen> _av500_: know off hand if you are using av_seek_frame or doing something special for flac?
[20:26:36] * Kovensky is playing flac with foobar2000 right now
[20:26:51] <mru> traitor
[20:26:52] <_av500_> mchinen: i use file size/duration as avg bitrate and seek with that
[20:27:21] <_av500_> mchinen: i can pastebin u ulgy hack tomorrow
[20:27:28] <_av500_> ping me at work
[20:27:39] <_av500_> but its rather trivial
[20:27:54] <_av500_> i guess it would be rejected as it is not "exact"
[20:28:01] <mchinen> _av500_: cool thx, it would help.
[20:29:22] <jai> hmm, people still prefer assignments inside if
[20:29:32] <Kovensky> http://www.vuvux.com/ lol
[20:29:43] <mru> jai: prefer over what?
[20:30:16] <jai> mru: well assignment in separate statement
[20:30:24] <jai> mru: i'm referring to the matroskadec patch
[20:30:27] <mru> jai: depends on context
[20:30:37] <elenril> it's m04ar 1337
[20:30:55] <jai> i thought it was a bit more clear, but i guess that's just e
[20:30:57] <jai> *me
[20:31:25] <jai> anyway, i have better mkv patches :)
[20:31:45] <elenril> ordered chapters? ;)
[20:32:26] <ramiro> mru: "mingw seems mostly dead"? not really...
[20:32:29] <jai> ah, no :) just a few changes to the parser so it can work with some broken files etc
[20:32:36] <ramiro> wbs: what errors are those?
[20:32:39] <mru> ramiro: so why don't they fix the bugs?
[20:32:45] <mru> and why is to friggin outdated?
[20:33:37] <ramiro> they suck at making new releases, that's true...
[20:33:59] <ramiro> mingw-w64 is becoming a great replacement, even for 32-bit code.
[20:34:15] <mru> then again, why would competence be expected from windows users?
[20:37:24] <ramiro> why am I in this channel anyways?
[20:37:35] <ramiro> wbs: feel free to pm me
[20:38:49] <janneg> that came unexpected
[20:39:03] <mru> when did he get that touchy?
[20:42:06] <Honoome> o-O and that was a mild remark, coming from mru
[20:44:42] <lu_zero> o_O
[20:45:02] <lu_zero> bad day got worse?
[20:45:48] * Honoome decides to work a bit on his autotools guide tonight
[20:46:14] * mru throws snide remark about autotools users at Honoome 
[20:46:14] * Kovensky libtoolizes Honoome
[20:46:26] <Honoome> mru: Kovensky got the upper hand, he wins :P
[20:46:49] * lu_zero throws requests for split ffmpeg configure in ffbuild
[20:46:52] * mru moves Honoome to another directory and watches him break
[20:46:53] <Honoome> btw, feng does everything without libtool!
[20:47:07] <lu_zero> and w/out recursion
[20:47:18] <Honoome> [until lu_zero gets his pony and we start to ship plugins
[20:47:30] <lu_zero> isn't my pony
[20:47:31] <Honoome> which I hope will be so late that dolt will be an acceptable solution]
[20:47:41] <Honoome> definitely is your pony
[20:47:46] <lu_zero> not mine
[20:47:53] * _av500_ thought lu_zero needs a llama
[20:47:54] <lu_zero> they throw it to me
[20:48:00] <lu_zero> _av500_: I have two!
[20:48:04] <lu_zero> and they are cute!
[20:48:08] <_av500_> right
[20:48:14] <mru> llamas?
[20:48:18] <Honoome> lu_zero: one is called "ciaran" and the other "daniel"?
[20:48:23] <Kovensky> again with the llama story? :P
[20:48:29] <lu_zero> http://lu-zero.deviantart.com/badges/
[20:48:42] <lu_zero> Honoome: those aren't llama
[20:48:51] <lu_zero> and I won't call them cute
[20:48:51] <Kovensky> derp, I'm supposed to be doing this boring PHP homework to deliver tomorrow, not chatting on IRC for 3h straight
[20:49:09] <mru> my sister has an actual, live llama
[20:49:34] <lu_zero> you are quite peculiar, so your sister.
[20:49:47] <mru> in very different ways
[20:50:24] <lu_zero> =)
[21:20:11] <elenril> siretart: does 0.6 support webm muxing?
[21:24:21] <_av500_> it should
[21:48:49] * peloverde just wasted 8 hours on an off-by-1 bug
[21:50:19] <peloverde> http://github.com/aconverse/ffmpeg-heaac/commit/f3ee85e2e6603637f2fd8d48be52a62946dcd7bc
[22:00:33] * Honoome wants to cry
[22:10:34] <hyc> I never got anything to build with mingw-64. it's actually usable now?
[22:10:54] <hyc> I guess I shuold look into it sometime, considering I'm on the mingw dev team...
[22:11:14] * mru blames hyc
[22:11:47] <hyc> hey, I was drafted after I submitted a bug fix
[22:12:24] <hyc> I was never responsible for release management ...
[22:13:59] <mru> funny, everybody says that
[22:14:22] <hyc> ;)
[22:16:22] <hyc> on a completely different topic, does anybody here have an Alpha server online? running TRU64?
[22:16:46] <mru> alpha yes, tru64 no
[22:17:06] <mru> been thinking of setting it up though
[22:17:11] <mru> just for laughs
[22:17:18] <hyc> you're running linux?
[22:17:21] <mru> yes
[22:17:29] <mru> on a up1500
[22:17:34] <mru> it can only run linux
[22:17:41] <mru> and maybe some bsd
[22:18:10] <hyc> I posted a patch to the tty driver to lkml, one of the feedbacks was that ioctl bit definitions for alpha linux have to be compatible with DEC
[22:18:20] <mru> yes
[22:18:28] <mru> so linux can run tru64 binaries
[22:18:32] <hyc> I was hoping to get a look at their <termios.h>
[22:18:41] <mru> ask in #alpha
[22:18:53] <hyc> but I tend to think this is not relevant, since this bit was previously not defined
[22:18:57] <hyc> ok, thx
[22:36:33] <bcoudurier> hi guys
[22:37:14] <BBB> hellow
[22:39:09] <CIA-92> ffmpeg: bcoudurier * r23620 /trunk/libavcodec/svq3.c: Fix svq3 watermark log uncompress on 64bit, probably fixes issue 1263
[22:43:24] <peloverde> As far as "development policy" goes, is HE-AACv2 considered a new codec?
[22:43:57] <hyc> bcoudurier: I still have a couple tiny patches on ffserver outstanding. e.g., bypass seek on a newly opened stream if stream_pos == 0
[22:44:13] <hyc> and that whole bit has been inside a #if 1 for a long time, shouldn't it be unifdef'd now?
[22:44:24] <hyc> (ffserver.c:2192 or so)
[22:46:15] <bcoudurier> peloverde, what are your concerns ?
[22:46:38] <bcoudurier> hyc, what's the problem with the seek ?
[22:46:56] <hyc> probably no problem now with the other timestamp fixes in
[22:47:07] <hyc> but why are we seeking to zero on a newly opened stream?
[22:48:35] <bcoudurier> humm I'm not sure
[22:48:44] <bcoudurier> but if it's there it was for a reason :>
[22:49:03] <peloverde> new codecs need thorough review before landing, new features in existing codecs only need maintainer's approval.
[22:49:05] <hyc> if they specified some other start position, sure that would be obvious
[22:49:32] <hyc> I guess I can drop that patch. but the #if 1 surrounding it always catches my attention too.
[22:49:53] <bcoudurier> peloverde, if you are confident enough, that should be fine
[22:51:19] <hyc> bcoudurier: and this one is more vital https://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2010-June/090008.html
[22:51:43] <mru> peloverde: anything major does well with a thorough review
[22:54:05] <bcoudurier> this one seems ok
[22:57:33] <hyc> bcoudurier: ok? then I'll commit that soon, thx
[23:11:00] <Honoome> lu_zero: around?


More information about the FFmpeg-devel-irc mailing list