[FFmpeg-devel-irc] IRC log for 2010-08-04

irc at mansr.com irc at mansr.com
Thu Aug 5 02:00:16 CEST 2010


[00:00:28] * J_Darnley sleeps
[00:05:16] <Dark_Shikari> J_Darnley: that only did a single one
[00:07:24] <Dark_Shikari> oh, nevermind, I missed a space
[00:09:08] <CIA-92> ffmpeg: darkshikari * r24688 /trunk/libavcodec/options.c: Cosmetics: add missing space to options.c
[00:49:04] <Dark_Shikari> heh.  if I use a while loop instead of a goto in decode_coeffs, gcc actually gets dumber
[00:49:21] <Dark_Shikari> e.g. for skip_eob
[01:14:43] <kierank> some people have been taking the fflame pill: http://www.bcs.org/server.php?show=conWebDoc.36428
[01:38:59] <CIA-92> ffmpeg: darkshikari * r24689 /trunk/libavcodec/vp8.c:
[01:39:00] <CIA-92> ffmpeg: VP8: simplify decode_block_coeffs to avoid having to track nonzero coeffs
[01:39:00] <CIA-92> ffmpeg: Slightly faster.
[01:40:14] <Dark_Shikari> oh fuck, I committed a start_timer
[01:41:46] <CIA-92> ffmpeg: darkshikari * r24690 /trunk/libavcodec/vp8.c:
[01:41:46] <CIA-92> ffmpeg: Fix 100L in r24689
[01:41:46] <CIA-92> ffmpeg: Accidentally committed some timing code.
[02:24:17] <CIA-92> ffmpeg: darkshikari * r24691 /trunk/libavcodec/vp8.c:
[02:24:17] <CIA-92> ffmpeg: VP8: partially inline decode_block_coeffs
[02:24:17] <CIA-92> ffmpeg: Avoids a function call in the case of empty DCT blocks (most of the time).
[03:06:49] <efriedma> Dark_Shikari: r24691 appears to be slower than r24688
[03:15:21] <efriedma> Dark_Shikari: meh, nevermind... the difference is within measurement error
[03:20:06] <Dark_Shikari> efriedma: 6 cycles faster here
[03:20:20] <Dark_Shikari> it's a tried-and-true strategy, used in x264 too
[03:26:13] <efriedma> Dark_Shikari: my results: http://paste.pocoo.org/show/245601/
[03:30:26] <Dark_Shikari> o.0
[03:30:40] <Dark_Shikari> I had consistent wins from both, at minimal stddev
[03:32:44] <efriedma> maybe it depends on the gcc version?
[03:32:57] <efriedma> a goto into a loop often messes with loop optimizers
[03:43:12] <Dark_Shikari> I used 4.3
[03:43:22] <Dark_Shikari> in my version, it doesn't compile a jmp at all
[03:43:33] <Dark_Shikari> so there's probably another way to construct the loop to avoid it
[03:44:38] <efriedma> i can try compiling with gcc 4.3... would that be of interest?
[03:45:23] <Dark_Shikari> dunno
[03:45:27] <Dark_Shikari> feel free to experiment
[03:58:18] <efriedma> hmm, strange... the times are almost exactly the same patched and unpatched with gcc 4.3
[03:58:25] <efriedma> and both are faster than gcc 4.4
[03:58:45] <efriedma> gcc's making wonderful progress :)
[03:59:19] <Dark_Shikari> lol
[03:59:23] <Dark_Shikari> it's the odd/even rule
[03:59:25] <Dark_Shikari> opposite of star trek
[03:59:28] <Dark_Shikari> odd == good, even == bad
[04:03:31] <peloverde> I always found the star trek rule to be ridiculous
[04:04:08] <Dark_Shikari> You're right.
[04:04:14] <Dark_Shikari> 10 sucked.
[04:04:38] <peloverde> 4 was way too campy
[04:05:08] <Dark_Shikari> That didn't make it less good.
[04:05:40] <peloverde> It wasn't any amount of good
[04:06:41] <Dark_Shikari> Yes, it was an unmeasurable amount of good.
[04:09:52] <peloverde> It was awful, it had whales, and spock wore that stupid headband, and the bit with scotty and the computer was one of the stupidest things i've ever seen
[04:10:13] <Dark_Shikari> you have no soul
[04:10:49] <efriedma> wow, and binary compiled with gcc trunk is like 7% faster overall than one compiled with gcc 4.4
[04:11:58] <Dark_Shikari> o.0
[04:11:59] <Dark_Shikari> 0.o
[04:12:04] <Dark_Shikari> how does that compare to 4.3?
[04:15:02] <efriedma> Dark_Shikari: 3138 dezicycles in decode_block_coeffs vs about 3500 with gcc 4.3
[04:15:59] <Dark_Shikari> holy shit
[04:16:09] <Dark_Shikari> so, how does it compare with and without patch?
[04:16:12] <Dark_Shikari> with trunk gcc
[04:29:42] <efriedma> Dark_Shikari: the difference is only a cycle or so
[04:30:06] <Dark_Shikari> ok, so it doesn't hurt at least.
[04:30:11] <Dark_Shikari> it might depend on the clip.
[04:36:07] <efriedma> i know... we should rewrite decode_block_coeffs_internal in asm to make it even faster :)
[05:05:53] <Dark_Shikari> so yeah, try benching with other clips
[05:05:59] <Dark_Shikari> maybe the effectiveness depends on the number of empty blocks
[05:06:09] <Dark_Shikari> of that patch
[05:06:50] <thresh> moroning
[05:21:43] <CIA-92> ffmpeg: mstorsjo * r24692 /trunk/libavcodec/Makefile:
[05:21:43] <CIA-92> ffmpeg: Fix VP8 decoder dependencies
[05:21:43] <CIA-92> ffmpeg: This reverts rev 24674 - the VP8 decoder actually depends on cabac.o.
[05:21:43] <CIA-92> ffmpeg: vp8.c includes vp56.h, which includes cabac.h, which has inline functions
[05:21:43] <CIA-92> ffmpeg: that reference tables from cabac.c.
[05:21:44] <CIA-92> ffmpeg: This fixes compilation with --disable-everything --enable-decoder=vp8.
[05:48:27] <lu_zero> good morning
[05:48:58] <lu_zero> http://26-26-54.hardwarebug.org/?id=28 <- the question is how
[05:49:13] <lu_zero> having a linmpeg or an ffnix?
[05:52:03] <lu_zero> efriedma: how gcc-4.5 and clang fare on that code?
[05:53:26] <efriedma> lu_zero: i don't have gcc-4.5 handy; clang is similar to gcc 4.3
[05:55:28] <efriedma> but clang is something like 10% slower than gcc 4.3 overall
[05:56:04] <efriedma> i think it has something to do with loop unrolling because the clang vp8.o is substantially smaller than the gcc one
[05:57:43] <Dark_Shikari> or inlining
[05:57:47] <Dark_Shikari> does av_always_inline work for clang?
[05:59:00] <astrange> biggest optimization differences: llvm inlines less (but sanely, i don't remember details), gcc's unroller runs way too often, llvm loop iv optimizer is worse, llvm register allocator is much worse than 4.4+
[06:00:14] <saintdev> peloverde: ping
[06:00:37] <Dark_Shikari> wow.  register allocation worse than gcc?
[06:02:14] <astrange> most free compilers are (because they use linear scan and aren't designed for x86)
[06:02:52] <astrange> probably icc, msvc, pathscale are better but i haven't tested them
[06:03:06] <mru> morning
[06:03:23] <lu_zero> I'm pondering about messing up with ack just to see how incomplete it is
[06:03:42] <lu_zero> but I'm afraid it's a pointless exercise
[06:03:49] <mru> astrange: otoh icc and pathscale miscompile ffmpeg, and msvc isn't a compiler at all
[06:03:57] <saintdev> lol
[06:04:16] <lu_zero> efriedma: I noticed that clang works and bootstraps on arm, you got any figure about it?
[06:05:08] <peloverde> saintdev: I'm barely still awake, shoot me an e-mail or ping me again tomorrow
[06:05:18] <efriedma> lu_zero: i don't have an arm device to test with
[06:05:50] <saintdev> peloverde: just wanted to request a "jump to block X" feature for aacx if there isn't one yet.
[06:06:07] <peloverde> There is, did you see the screeshot?
[06:06:27] <saintdev> no, only the actual mp3x screenshots, and i've never used it before :P
[06:06:46] <saintdev> ok, cool. that's all i wanted ;)
[06:07:00] <peloverde> http://i.imgur.com/WfVP8.png
[06:07:04] <saintdev> at least for now :)
[06:07:30] <saintdev> cool :D
[06:07:44] <peloverde> I will add window overlays soon, and make the scalefactors accurately sized
[06:08:01] <saintdev> how about zoom?
[06:08:35] <peloverde> Right now the default zoom level is one frame, it will probably remain so for the time being
[06:08:46] <peloverde> default=only
[06:09:08] <astrange> i'd like to add compiler benchmarks to fate but don't want to write boxplot drawing
[06:09:39] <lu_zero> efriedma: I'll try soon =)
[06:09:39] <saintdev> ok
[06:10:10] <lu_zero> btw the clang and llvm HEAD now need to be built out of tree, isn't it?
[06:10:58] <peloverde> fate seems awfully yellow these days
[06:11:43] <mru> it's my sheevaplug having a meltdown
[06:11:52] <mru> I've inserted sleeps in the testing to let it cool off
[06:12:19] <mru> that thing is the shittiest hw I've seen
[06:13:19] <mru> it's also more yellow due to all the new tests
[06:13:23] <peloverde> true
[06:13:26] <mru> which fail here and there
[06:13:57] <peloverde> pathscale isn't doing that great, I though pathscale was the real future and llvm was just hype (tm)
[06:14:20] <mru> they say they'll fix it some time
[06:14:34] <astrange> llvm's corporate sponsorship really needs a system compiler
[06:14:46] * peloverde feels vindicated for never drinking the pathscale coolaid... and sleepy
[06:15:05] * mru never believes anything until he sees it
[06:15:51] <mru> those vorbis failures on bsd are curious
[06:16:53] <saintdev> mru: nice bar at the top ;)
[06:17:06] <peloverde> vorbis fails a lot of places, aac and wma also fail quite a few places
[06:17:22] <peloverde> I wonder if the imdct/ifft is getting miscompiled
[06:18:00] <astrange> too bad about the no unit testing
[06:18:09] <saintdev> mru: next suggestion, separate the bar into individual slots
[06:18:10] <mru> let's add fft-test
[06:18:13] <astrange> is it possible for the imdct to cause bitstream errors like that?
[06:18:25] <mru> astrange: I doubt it
[06:18:30] <mru> not even xiph are that mad
[06:18:41] <lu_zero> open64 for amd64 from amd...
[06:18:50] <lu_zero> yet another compiler to try ^^
[06:18:53] <lu_zero> brb
[06:18:54] <peloverde> you guys are right
[06:18:59] <peloverde> still fft-test would help
[06:19:58] <saintdev> peloverde: weren't you going to bed :P
[06:20:11] <peloverde> I am very soone
[06:20:26] <peloverde> but I can never resist poking fun at the *BSD
[06:20:40] <mru> netbsd works somehow
[06:21:27] <mru> I need to add a way of showing which setup is running valgrind
[06:22:21] <saintdev> "(valgrind)"
[06:22:34] <mru> sure, but there's no way of adding to any of the fields
[06:22:46] <mru> so I need a new one
[06:24:19] <mru> and gcc 4.5 on x86_32 should be investigated
[06:24:49] <peloverde> yes
[06:25:06] <mru> mxf demuxer and nut muxer are broken
[06:25:28] <peloverde> I wish google would run fate with whatever wonky mingw fork they use, though they probably have some sort of liablity excuse as to why they can't
[06:27:16] <lu_zero> mru: gcc had and has its share of regressions
[06:27:29] <mru> I know that
[06:27:31] <lu_zero> peloverde: let's ask
[06:28:04] <saintdev> peloverde: they don't use ffmpeg. wink, wink.
[06:28:16] <peloverde> I'm talking about chrome here
[06:28:45] <saintdev> oh
[06:57:45] <CIA-92> ffmpeg: mru * r24693 /trunk/libavcodec/fft-test.c: fft-test: exit with non-zero status if test failed
[06:57:45] <CIA-92> ffmpeg: mru * r24694 /trunk/tests/fate-run.sh: fate-run: add helper to run things on target
[06:57:46] <CIA-92> ffmpeg: mru * r24695 /trunk/tests/fate2.mak: fate: add fft tests
[06:58:09] <mru> there, let's see what fails
[07:04:26] <ohsix> Dark_Shikari: what was that 2 letter tld url shortener?
[07:04:53] <_mru_> to
[07:05:29] <_mru_> but it's not working
[07:06:24] <ohsix> hm thats wat i thought, thanks
[07:12:39] <pJok> is it me or does fate lack windows builds?
[07:12:49] <mru> it does
[07:12:59] <mru> guess we're waiting for ramiro
[07:13:06] <kshishkov> or you :P
[07:13:16] <wbs> are there instructions anywhere on how to set up a machine to run this new fate?
[07:14:44] <mru> https://lists.mplayerhq.hu/pipermail/fate/2010-July/000260.html
[07:15:33] <mru> and there's a typo in the svn url
[07:16:24] <wbs> ah, thanks, I didn't know of that list. and the announcment on -devel felt a bit bare :-)
[07:16:59] <kshishkov> it's the ML for people with first names starting with 'M' anyway
[07:17:14] <kshishkov> so you fit :)
[07:17:52] * KotH will stay away of that list
[07:18:45] * kshishkov too - unsuitable first name
[07:19:23] <wbs> mru: is there any interest in running compile-only test setups?
[07:26:10] <mru> hey look, suncc fails imdct
[07:28:03] <kshishkov> does anyone care about oraclecc?
[07:28:22] <kshishkov> though that's a shameful bug
[07:28:43] <kshishkov> even icc works better
[07:29:01] <KotH> even path64 works better :)
[07:30:00] <KotH> though path64 seems to miscompile something in the h264 code
[07:30:21] * kshishkov wonders why lavc H.264 is not the best one
[07:30:39] <KotH> because you didnt optimize it
[07:30:52] <cartman> because others are exploiting multicore shit
[07:33:22] * elenril wonders if -mt will be merged before or after DNF
[07:33:53] <kshishkov> KotH: the problem lies in basic code, I think
[07:34:14] <astrange> it's quite good now
[07:34:20] <KotH> then rewrite the basic code... but please not in basic
[07:34:33] <kshishkov> cobol?
[07:34:51] <kshishkov> astrange: how good it's compared to competitors?
[07:34:51] <elenril> i've heard c# is teh new shit
[07:34:56] <astrange> some branches are left to remove from high level code, hl_motion is too big, loopfilter is too big, every data structure is too big
[07:35:07] <astrange> dunno, i don't have a way to benchmark anything on windows
[07:35:09] <kshishkov> c# is finally mutated M$ Java
[07:36:38] <mru> cartman: ffh264 is embarrassingly slow compared to others on single-core too
[07:37:10] <cartman> mru: gpu acceleration helps on windows at least
[07:37:13] <cartman> and Linux probably
[07:37:17] <mru> cartman: that's irrelevant
[07:37:26] <mru> that's a different decoder for all intents and purposes
[07:37:28] <kshishkov> and I suspect that, say, CoreAVC and Nero H.264 decoder devs are not better than our devs either
[07:37:30] <cartman> in the end I don't care its fast for me ;)
[07:37:47] <astrange> commercial devs don't mind writing huge copy-paste decoders
[07:38:01] <mru> astrange: that isn't the case here
[07:38:15] <mru> I've looked at both
[07:38:59] <av500> mru has seen it all
[07:39:38] <cartman> mru is tainted :P
[07:39:49] <kshishkov> and cartman is doomed
[07:39:58] <cartman> who isn't doomed anyway
[07:40:01] <cartman> we are all gonna die :P
[07:40:17] <ohsix> we're dying right now
[07:40:22] <av500> omg
[07:40:25] * av500 runs
[07:40:30] * thresh definitely is
[07:40:30] * kshishkov writes to Turkish government that cartman wrote a joke about Mustafa Kemal
[07:40:31] <ohsix> slowly
[07:40:42] <kshishkov> that's called "living"
[07:40:44] <cartman> kshishkov: you learn too fast
[07:40:45] <av500> thresh: use less wodka
[07:40:53] <KotH> kshishkov: that will get him into prison, not killed
[07:40:57] <thresh> av500: one cant breathe, let alone drink vodka here
[07:41:06] <kshishkov> av500: but alcohol preserves living matter quite good
[07:41:08] <ohsix> thats wear and tear not covered by the limited warranty
[07:41:10] * cartman is not on VPN
[07:41:14] <cartman> won't make government jokes
[07:41:15] <cartman> :p
[07:41:25] <av500> Kemal wrote what part of ffmpeg?
[07:41:27] <kshishkov> KotH: I'll add some stuff about defending Armenians too
[07:42:06] <KotH> kshishkov: most turkish people do not care about the armenians
[07:42:07] <kshishkov> av500: no, he made Turkey more civilised country than it was in XIX century
[07:42:20] <cartman> just one word, Taqiyya, http://en.wikipedia.org/wiki/Taqiyya
[07:42:26] * av500 lives in XML century ...
[07:42:33] <KotH> kshishkov: they only care about other people thinking they know what happend 90 years ago and blame turkish people for it ;)
[07:42:50] * cartman blames the government for everything now
[07:43:17] * av500 looks at the nice web award logo Taqiyya got
[07:43:19] <KotH> cartman: s/goverment/mafia/
[07:43:31] <cartman> they are now hand in hand
[07:43:45] <kshishkov> they always were and they always will
[07:43:59] <KotH> hand in hand? i doubt that any respectable mafia boss would shake that goverments hand
[07:44:06] <cartman> lol
[07:44:19] <kshishkov> KotH: why? it's usually his man there
[07:44:42] <KotH> kshishkov: i've never seen anyone who's so shameles as these guys
[07:45:06] <kshishkov> "these guys"? who?
[07:45:15] <KotH> the current turkish "goverment"
[07:45:40] <cartman> since all the IRC etc is logged lets talk about something else
[07:45:42] <cartman> shall we
[07:45:56] <kshishkov> too late :)
[07:46:26] <cartman> I didn't say nothing :P
[07:46:45] <kshishkov> double negative cancels itself
[07:47:17] * KotH doesnt fear these shitheads
[07:47:32] <cartman> KotH: yeah good luck with that in .tr
[07:47:38] <KotH> worst they could do is put me in jail
[07:47:52] <KotH> they're not man enough to kill me
[07:48:06] <kshishkov> death is cheap, prison is more horrible
[07:48:11] <cartman> jail is now full of high profile guys too
[07:48:15] <cartman> you wouldn't be bored
[07:48:20] <KotH> yeah...
[07:48:21] <mru> KotH: are you a swiss citizen?
[07:48:30] <mru> yes of course you are
[07:48:42] <mru> wouldn't be doing army service otherwise
[07:48:47] <KotH> and we could finally make those plans to overthrow the goverment, which they are accusing all the people to have
[07:49:05] <KotH> mru: and i have a rifle under my bed
[07:49:05] <cartman> that'll happen in elections
[07:49:09] <cartman> lol
[07:49:35] <KotH> do you really think that the elections will get anywhere?
[07:49:48] <cartman> sure
[07:49:53] <cartman> or I am moving to Europe :P
[07:49:58] <cartman> works for me :P
[07:50:27] <KotH> they've twisted the election rules so much, that they cannot be outvoted... not with such an high percentage of people who think that voting for a religious party will buy them a pass to heaven
[07:50:57] <KotH> er.. sorry... "religious"
[07:51:00] <cartman> KotH: since I think you can read Turkish, see http://haber.gazetevatan.com/haberdetay.asp?Newsid=320975
[07:56:24] <KotH> fun...
[08:01:17] <cartman> http://www.informationclearinghouse.info/article11918.htm explains wtf is going on
[08:06:29] <mru> isn't that just another conspiracy website?
[08:10:27] <av500> s/conspiracy //
[08:11:43] <cartman> mru: no
[08:11:56] <cartman> well at least that article is not conspiracy
[08:11:59] <mru> it sure looks like one
[08:12:03] <mru> the website in general
[08:12:06] <cartman> its a well known project
[08:12:14] <mru> many conspiracy theories are
[08:12:28] <cartman> I see that "conspiracy" theory live here :)
[08:12:38] <mru> I'm just sceptical when _one_ person claims to have all this secret information
[08:12:47] <cartman> government didn't deny it
[08:12:54] <cartman> involvement in the great middle east project
[08:12:57] <mru> ok, maybe there's truth in this story
[08:13:01] <kshishkov> yes, like US plan to dumb Russia - which works without any effort from US side
[08:13:15] <cartman> they snoop all the telephone lines
[08:13:16] <mru> but the way that article is writting, it sets off all my nutcase warning bells
[08:13:17] <cartman> the internet
[08:13:18] <cartman> cell phones
[08:13:23] <mru> written
[08:13:32] <cartman> won't happen without big technological support
[08:13:37] <cartman> guess who have that?
[08:13:55] <kshishkov> Turkish Telecom?
[08:14:10] <cartman> kshishkov: TT can't snoop on you talking in a room
[08:15:20] <cartman> who provides top secret visuals from unmanned aircrafts in Turkey, maybe Julian Assange is in Turkey
[08:15:23] <cartman> who knows :P
[08:16:20] * kshishkov does not care about the country that still hasn't forgotten it was Ottoman Empire
[08:26:24] <cartman> kshishkov: we are not an empire, :)
[08:26:32] <cartman> KotH: http://haber.gazetevatan.com/basbakana-3-esli-danisman/321097/1/Gundem have a good laugh my friend
[08:26:55] <kshishkov> cartman: well, I still cannot find Kurdistan on the map
[08:27:06] <cartman> kshishkov: soon enough you will
[08:27:06] <av500> ...Upgrade mistress, his wife is 11 years old, explaining the love...
[08:27:21] * av500 loves gg translate
[08:27:55] <cartman> kshishkov: never forget Mustafa Kemal Ataturk's famous words; "Peace at home, peace in world"
[08:28:17] <mru> there's some wisdom there
[08:28:25] <kshishkov> cartman: never heard so will never forget
[08:28:32] <cartman> kshishkov: now you heard
[08:28:40] <cartman> Obama needs to read about him
[08:28:45] <mru> and what has been heard cannot be unheard
[08:29:05] <kshishkov> not in politics
[08:29:10] <av500> kshishkov: it is wild! http://media.photobucket.com/image/durch wilde kurdistan/piktscha-maker/DurchswildeKurdistan-1.jpg
[08:29:42] <cartman> thats not politics
[08:29:45] <cartman> thats basic humanity
[08:30:11] <av500> oops: http://i361.photobucket.com/albums/oo55/piktscha-maker/DurchswildeKurdistan-1.jpg
[08:30:28] <cartman> I shall tell RMS that Kurdistan pirates GNU/Linux
[08:30:32] <cartman> that'll teach them
[08:30:49] <kshishkov> av500: I've heard about that author. Weren't DDR producing films about Indians after his books (with some guy from Yugoslavia as main star)?
[08:30:51] <mru> no, tell him they refuse to say GNU/
[08:31:01] <mru> _that_ will really teach them
[08:31:02] <cartman> or that
[08:31:03] <cartman> :D
[08:31:09] <av500> kshishkov: DDR? BRD!
[08:32:25] <kshishkov> av500: since those films were known is USSR I thought it would be rather Socialistic part
[08:32:25] <av500> and yes, Gojko Mitic ftw
[08:32:45] <kshishkov> I *knew* that name
[08:34:43] <av500> kshishkov: how can one not know
[08:36:20] <kshishkov> ask any American
[08:37:42] <av500> kshishkov: btw, you can go see it live: http://www.karl-may-spiele.de/frontend/index.php
[08:44:17] <KotH> cartman: rotfl...
[08:44:24] <KotH> cartman: suits them well :)
[08:44:37] <cartman> yeah I need 2 extra wifes too
[08:44:42] <cartman> religion wants me to do so
[08:45:23] <mru> wives are overrated
[08:46:03] <KotH> mru: only make sure you get a low maintenance one
[08:46:08] <av500> does ffmpeg compile on illumos?
[08:46:17] <KotH> cartman: be carefull what you wish for
[08:46:22] <mru> av500: wtf is that?
[08:46:24] <cartman> true that
[08:46:27] <KotH> cartman: you might end up with 2 extra wifes ;)
[08:46:27] <cartman> 1 is already too many :P
[08:46:31] <av500> mru: live under a rock as well?
[08:46:40] <av500> http://www.illumos.org/
[08:46:58] <mru> is it another bsd fork?
[08:46:59] <KotH> cartman: lol
[08:47:07] <KotH> cartman: i'll tell her that when i meet her ;)
[08:47:12] <cartman> mru: openslowaris fork
[08:47:12] <av500> mru: no opensunos fork
[08:47:14] <cartman> KotH: lol
[08:47:20] <cartman> she'd kill me for joking :P
[08:47:29] <KotH> i know :)
[08:47:30] <mru> it's not solaris anymore, it's so larry's now
[08:47:51] <mru> cartman: see, that's the problem with wives
[08:48:16] <KotH> mru: and cartman has a _turkish_ girl as a wife too!
[08:48:31] <mru> I don't know what they're like
[08:49:09] * kshishkov once spent >six hours in a bus with Turkish woman and two Dutch people
[08:49:16] <mru> av500: geez, can't they put up a _text_ announcement?
[08:49:17] <KotH> well, compared to those, any other girl i've seen is docile
[08:49:19] <kshishkov> nothing special
[08:49:29] <mru> KotH: "the other"?
[08:49:52] <kshishkov> mru: the _only_ other girl he'd ever seen
[08:49:56] <KotH> ^^'
[08:50:08] <mru> are girls that rare in .ch?
[08:50:15] <KotH> stop mocking me! mock cartman!
[08:50:21] * mru mocks cartman 
[08:50:30] <KotH> well.. good looking girls are quite rare in .ch
[08:50:36] <KotH> at least in the german speaking part
[08:50:45] <KotH> and those who are good looking, aren't swiss
[08:51:02] <mru> it's funny how people everywhere say that about their local girls
[08:51:25] * kshishkov wonders what they say in Sweden
[08:51:33] <av500> mru: why? afraid your phones pdf reader is vulnerable?
[08:51:38] <KotH> mru: well... i consider the turkish, italian and yugoslavian girls who live here as locals ;)
[08:52:03] <mru> kshishkov: swedish girls are overrated imo
[08:52:18] * KotH agrees with mru 
[08:52:37] <kshishkov> mru: just point me to any non-overrated thing
[08:52:38] <KotH> but that might be because i dont have a thing for cheese white blond girls
[08:52:56] <kshishkov> mmm... cheese
[08:52:58] <cartman> mru: marriage is overrated
[08:53:04] <cartman> lots of girls in .tr :P
[08:53:07] <mru> KotH: and then you're not even considering personality
[08:53:22] * kshishkov wants a kilo of Västerbotten
[08:53:24] <mru> they're all either dumb blondes or rabid feminists
[08:53:40] <KotH> rabbit?
[08:53:54] <mru> I'm not joking
[08:55:36] <kshishkov> majority of Ukrainian girls I know can be compared to cows - in intellect and beauty
[08:55:45] <KotH> mru: you dont want to consider the personality of swiss girls... you really dont want to
[08:55:49] <cartman> rabid |ˈrabəd; ˈrā-|
[08:55:49] <cartman> adjective
[08:55:49] <cartman> 1 having or proceeding from an extreme or fanatical support of or belief in something : a rabid feminist.
[08:55:57] <cartman> I suspect mru has a dictionary!
[08:56:10] <av500> cartman: I suspect he swallowed one as a kid
[08:56:12] <KotH> cartman: mru doesnt need one
[08:56:14] <cartman> av500: lol
[08:56:32] <av500> he was still hungry after the C standard
[08:59:52] <KotH> av500: and as we can see, neither c standards nor dictionaries are very nutricious...
[09:00:27] <mru> "nutritios"
[09:00:37] <av500> lol
[09:00:46] <mru> derived from nutrient, therefor spelt with a t
[09:00:54] <cartman> see
[09:01:04] <KotH> thanks
[09:01:33] * av500 wonders who this Ruby is that Honoome keeps blogging about
[09:01:34] <mru> compare "suspicious", derived from "suspicion"
[09:02:06] <cartman> av500: Ruby Goldberg
[09:02:08] <kshishkov> av500: a precious gem or not-so-precious OO language
[09:02:19] <mru> could be a girl
[09:02:31] <mru> how does one hack a girl?
[09:02:38] <av500> well, he does write her a lot of "love" letters
[09:02:42] <kshishkov> with rubies, of course!
[09:03:10] <kshishkov> av500: love was inventer in order to save some money
[09:03:15] <kshishkov> *invented
[09:04:13] <KotH> you mean girls invented love to save their money?
[09:04:21] <microchip_> kshishkov: lol?
[09:04:40] <mru> "Certainly there are lots of things in life that money won't buy, but it's very funny -- Have you ever tried to buy them without money?" -- Ogden Nash
[09:04:49] <kshishkov> KotH: men did, obviously. To prevent girls from draining all their money.
[09:05:00] <mru> kshishkov: the plot failed
[09:05:02] <microchip_> kshishkov: double lol?
[09:05:22] <kshishkov> mru: obviously
[09:05:34] <mru> I'd even say it backfired
[09:06:03] <kshishkov> could be
[09:10:32] <KotH> mru: "Money doesn't make you happy, but it simplifies a lot of things" - Nejat Kinali
[09:10:56] <mru> that is true
[09:11:13] <kshishkov> KotH: Money can't buy you happiness but you can rent it with them. - Don't remember who
[09:11:27] <microchip_> KotH: depends on the person. Some are fully happy by having only lots of $$$, they couldn't care less for other things ;)
[09:11:50] <KotH> microchip_: yeah.. i know such a guy
[09:11:53] <mru> microchip_: miserable people
[09:11:56] <microchip_> horrible :/
[09:11:59] <mru> annoying as hell too
[09:12:02] <microchip_> yep ^^^
[09:12:03] <kshishkov> I know that tax office too
[09:12:13] <KotH> microchip_: but i dont think he's happy with it either... he always tries to get rid of all the money he has
[09:12:26] <mru> KotH: I'll help him
[09:13:08] * KotH remembers how he bought a 200chf gold necklage for the girl he loved... and slapped her, when she rejected it...
[09:13:52] <kshishkov> that's unacceptable
[09:13:52] <av500> with the necklace?
[09:13:56] <mru> don't buy stuff for girls
[09:14:05] <kshishkov> you should never slap girl - only kill her
[09:14:10] <KotH> lol
[09:14:15] <mru> the moment you start buying stuff they'll just string you along for the money
[09:14:23] <kshishkov> (heard at the university)
[09:14:27] <KotH> common boys, he didnt know any better.. he was just 12!
[09:14:28] <microchip_> kshishkov: first rape, then kill :P
[09:15:47] <kshishkov> KotH: then you've been much wealthier than me then. Those were absolutely incredible money for me >10 years ago
[09:16:25] <KotH> kshishkov: me? i was happy to have 2chf in my pocket back than!
[09:16:42] <KotH> kshishkov: as a comparison 200chf were about 300DM (no euro yet)
[09:16:46] <mru> now 200chf is pocket money
[09:16:54] <av500> KotH: get trousers without holes
[09:17:13] <KotH> av500: would have to go to .tr first
[09:17:21] <av500> you spend on the trousers, but you keep more money
[09:17:51] <kshishkov> KotH: well, even 50DM were quite incredible back then. And I rarely got more
[09:17:58] <kshishkov> more than 2chf
[09:18:11] <KotH> kshishkov: you sound like ods15
[09:18:32] <KotH> "that glass of water was just 3 months salary"
[09:18:36] <cartman> I remember our house was 45000 DM
[09:18:54] <KotH> cartman: was it still standing in 2000? ;-)
[09:18:57] <av500> ye olde times
[09:19:01] <cartman> KotH: sure
[09:19:19] <cartman> it was built in 99 I think
[09:19:24] <cartman> was way too expensive
[09:19:34] <kshishkov> KotH: Ukraine is like Turkey, so you could live with laughably small amount of European money then
[09:19:38] <KotH> before or after the quake?
[09:19:44] <cartman> KotH: before
[09:20:39] <cartman> kshishkov: why the girls in Ukraine are just too hot? // this must be asked! :p
[09:21:15] <microchip_> cartman: is your first name Eric? ;)
[09:21:22] <cartman> microchip_: maybe :P
[09:21:30] <microchip_> hehe
[09:21:30] <kshishkov> cartman: because of the weather - it's around 40 degrees nowadays there
[09:21:43] <cartman> kshishkov: always funny you are
[09:22:37] <microchip_> cartman: this you? http://3.bp.blogspot.com/_oW0XBl9qmGo/SwuitD5qHlI/AAAAAAAAAIg/iT94FmDQ2KE/s1600/eric-cartman.jpg
[09:22:38] <kshishkov> told you, most of Ukrainian girls are as smart and beautiful as cows
[09:22:54] <cartman> microchip_: lol
[09:22:57] <cartman> kshishkov: well
[09:23:01] <cartman> maybe its the hype
[09:23:02] <cartman> ok
[09:23:03] <cartman> :)
[09:23:10] <cartman> microchip_: thats an evil look
[09:23:14] <microchip_> ;)
[09:23:23] <microchip_> you ARE evil! :p
[09:23:37] * cartman is barely evil
[09:23:42] <microchip_> and you should go annoy that jew instead of hanging around on irc ;)
[09:23:46] <KotH> cartman: i thought that the ru/ua-girl hype was in early 2000s?
[09:23:51] <kshishkov> cartman: and probably if she has good look she will use it to catch a husband and not work anymore in her life
[09:23:55] <cartman> KotH: still going on strong
[09:24:03] <cartman> kshishkov: that still happens here :D
[09:24:07] * KotH shakes his head
[09:24:13] <microchip_> kshishkov: what about Mila Jovovitch?
[09:24:21] <KotH> cartman: and  that, although you've so many hot girls!
[09:24:21] <cartman> -t
[09:24:22] <mru> overrated
[09:24:30] <mru> microchip_: ^^
[09:24:34] <cartman> KotH: ah
[09:24:42] <cartman> thank god we have pr0n anyway
[09:24:47] <KotH> lol
[09:24:48] <kshishkov> microchip_: please return when you have at least a thousand of examples
[09:25:00] <cartman> salutes to Arpi for bringing pr0n to Linux
[09:25:13] <microchip_> kshishkov: same thing can be said about the girls in most other countries
[09:25:35] <KotH> cartman: rotfl
[09:25:51] <microchip_> cartman: don't forget the animu for KotH :p
[09:25:53] <kshishkov> cartman: in my university days there were about a dozen of girls in our group. Most of them were neither good looking nor smart enough. And at least quarter of them married working idiots
[09:25:53] * microchip_ runs
[09:25:58] <cartman> "ukraine girls" in Google images
[09:26:00] <cartman> shows pr0n
[09:26:11] <cartman> kshishkov: working is good
[09:26:13] <KotH> cartman: you know, that beside arpi and a few other poor guys, most of the mplayer devels of the early days used mplayer to watch anime? :)
[09:26:33] <cartman> KotH: yeah hentai!
[09:27:00] <microchip_> bring back Gabu!
[09:27:06] <cartman> lol
[09:27:08] <kshishkov> cartman: yes, especially if it's not you who works and you get a loin's share of his income
[09:27:09] <cartman> oh yeah
[09:27:25] <cartman> kshishkov: I hope the sex is worth the money then!
[09:27:28] <cartman> ;)
[09:27:28] <kshishkov> KotH: I used MpegTV player
[09:27:43] <cartman> I used some shit that only decodes mpeg2
[09:27:44] <microchip_> KotH: is Richard Felker still active?
[09:27:47] <KotH> microchip_: thanks but i'm not into yaoi
[09:27:51] <av500> a loin's share, hmm....
[09:27:52] <microchip_> lol
[09:27:58] <KotH> microchip_: for low values of active
[09:28:14] <cartman> ah Rich Felker the flametastic guy
[09:28:24] <cartman> should run for president in .tr
[09:28:27] <kshishkov> av500: sorry, race condition. it was intented to be "lion's"
[09:28:29] <microchip_> lol
[09:28:50] <kshishkov> cartman: just import current president from Iran
[09:29:04] <cartman> I should compile mplayer with gcc 2.96 and send Felker a screenshot
[09:29:05] <cartman> that'll do
[09:29:07] <microchip_> epic loony tunes :D
[09:29:56] <cartman> kshishkov: Mr. Ahmediwhoknowstherest
[09:30:32] <kshishkov> cartman: do you knwo the name of current Turkmenistand god^W president?
[09:30:43] <cartman> nope
[09:30:52] <cartman> not me, all I can say
[09:30:53] <microchip_> haraparashorashkov?
[09:30:56] <kshishkov> Gurbanguly Berdimuhamedow
[09:31:02] <cartman> lol
[09:31:31] <cartman> Erican Cartmanow
[09:31:39] <cartman> you should be picky about the names when electing
[09:33:03] * microchip_ would have found it extremely lulzy if Barak Obama was called Barak Osama and was elected US president with that name :D
[09:33:08] <kshishkov> since you elect names, yes
[09:33:26] <cartman> microchip_: lol
[09:33:27] <kshishkov> microchip_: just remember hist second name
[09:33:30] <microchip_> it's just one letter change!
[09:33:32] <cartman> Hussain
[09:33:36] <microchip_> yeah
[09:33:44] <av500> Saddam Osama
[09:33:49] <cartman> Barack rhymes with dick in .tr
[09:33:50] <cartman> :P
[09:33:59] <microchip_> Satan Hussein Osama :D
[09:34:06] <cartman> so I knew he would screw us
[09:34:30] <kshishkov> why? He got nice military first name?
[09:35:16] <microchip_> in Bulgarian, Barak(a) means a shack :D
[09:35:35] <cartman> kshishkov: no his name rhymes with male body parts
[09:35:37] <kshishkov> it does so in many languages, including English
[09:35:42] <cartman> he's the man of his name
[09:35:58] <mru> microchip_: in swedish too
[09:36:08] <kshishkov> microchip_: and in USSR it was actually a name for living house as well
[09:36:22] <microchip_> mru: he has a nice shack though, the white house :O
[09:36:30] <kshishkov> black house
[09:36:40] <av500> black barakc
[09:38:30] <kshishkov> yes, that's how it's often called now in some part of world far from US
[09:39:06] <microchip_> kshishkov: is russia still burning now? fire/smokes still going on?
[09:39:25] <microchip_> i saw pics of moscow, full of smoke could barely see through
[09:39:42] <kshishkov> microchip_: dunno. And I never cared. Ask thresh when he's around
[09:39:49] <microchip_> k
[09:39:54] <av500> he had internet access one hour ago...
[09:43:41] <KotH> when did this channel turn into #ffpolitics?
[09:43:51] <KotH> oh.. where are the days when it was still #ffanime
[09:44:04] <av500> it was a lot of #ffgirls as well
[09:44:08] <av500> (today)
[09:44:27] <kshishkov> KotH: it was called #mplayer-dev then :P
[09:44:55] * mru there are no ffgirls
[09:45:12] <av500> #ffaboutgirls then
[09:45:13] <KotH> mru: apropos, what happend to the ffgirl?
[09:45:24] * kshishkov looks at http://hardwarebug.org/wp-content/uploads/2010/02/img_3133.jpg
[09:45:47] <av500> kshishkov: the one with the long hair?
[09:46:21] <KotH> av500: that's carl eugen ;)
[09:46:22] <kshishkov> av500: I try to spot you on that picture as well
[09:46:31] <KotH> av500: or to you mean luca? ;)
[09:46:37] <microchip_> mru: there is #netbsd-furries ;)
[09:46:37] <cartman> kshishkov: who is who?
[09:46:43] <cartman> I demand a response
[09:46:54] <KotH> cartman: the turkish looking guy is me :)
[09:47:12] <mru> the one with the shotgun is me
[09:47:13] <microchip_> KotH: with the beard?
[09:47:13] <av500> the av500 looking guy is me
[09:47:13] <cartman> KotH: there is a Turkish looking guy there? :D
[09:47:21] <KotH> mru: lol, i was about to say that :)
[09:47:51] <cartman> no who-is-who ?
[09:48:03] <kshishkov> actually I used it to train people  here- "that guy to the right is MÃ¥ns, the rest of folk are not significant"
[09:48:05] <microchip_> who's the chick?
[09:48:13] <av500> Diego
[09:48:15] <KotH> cartman: come to lt,fosdem,froscon and learn the names of the people there :)
[09:48:23] <cartman> KotH: would love to
[09:48:24] <microchip_> lol ^^
[09:48:28] <cartman> maybe someday
[09:48:35] <thresh> aboot?
[09:48:41] <KotH> cartman: bah..
[09:48:51] <KotH> cartman: alternatively, invite ffpeople to .tr
[09:49:01] <cartman> KotH: I am trying to get outside of .tr
[09:49:09] <cartman> ffpeople are welcome here, that can be arranged :P
[09:49:55] <cartman> a free multimedia conference like thinfg
[09:49:58] <cartman> could be done
[09:50:02] <av500> cartman: /me is the largest ffmpeg contributor
[09:50:04] <benoit-> mru: the shotgun was in your right hand, we cannot see it!
[09:50:15] <cartman> av500: yeah I know, you pollute git log :P
[09:50:15] <KotH> av500: rotfl!
[09:50:15] * kshishkov won't go to Turkey willingly - too many Russians there
[09:50:29] <kshishkov> KotH: but that's true
[09:50:31] <cartman> kshishkov: we over-welcome visitors :P
[09:50:33] <KotH> kshishkov: the russians are in the south
[09:50:38] <cartman> especially free software guys
[09:50:47] <KotH> kshishkov: busy stealing toilets
[09:50:50] <cartman> LOL
[09:50:53] <cartman> hahah
[09:51:02] <KotH> cartman: that's no joke
[09:51:09] <cartman> still funny damn it
[09:51:31] <KotH> cartman: we heard quite a few stories in kas last year
[09:51:34] <kshishkov> Turkish police report: last week there were 3 privies, 2 outhouses, 50 WCs and 1 loo stolen
[09:51:38] <cartman> :D
[09:52:47] <KotH> looks like the russians come to turkey, drink until they cannot even lift a finger and when they go back, they take everything in the hotel room that they can cary
[09:52:49] <kshishkov> "unknown burglars broke in, sawed off half of the bathroom and went away"
[09:53:03] <cartman> mv kshishkov seinfeld
[09:53:42] <kshishkov> KotH: yes, that sounds perfectly valid. Just add "sometimes they drink and grill themselves at the beaches"
[09:56:17] * av500 has still not done an inch of work today, blames #ffmpeg-devel
[09:56:53] * mru has pasted several hundred lines of code from one file to another
[09:57:04] * kshishkov has never measured hios work in inches, blames living in metric country and having non-linear work
[09:57:42] <cartman> I am supposed to write a video player, I am in the right place:P
[09:59:19] <av500> join #ffplay-devel
[10:00:09] <cartman> I use vlc :P
[10:02:22] <mru> cartman: there's #videolan for that
[10:02:29] <cartman> I am there too
[10:02:32] <mru> it's also full of trolls
[10:02:37] <cartman> I am discussing low level details here
[10:02:39] <cartman> right
[10:02:39] <cartman> :P
[10:02:55] <mru> av500: set your display to a lower dpi, you'll get more inches per line of code
[10:03:06] <mru> morning saste
[10:03:08] * cartman sets font size to 5px
[10:03:20] <mru> 5x7 is a good font size
[10:03:38] <cartman> not without my glasses or a 30'' screen
[10:03:40] <kshishkov> 1x1 is better
[10:04:15] <saste> mru: hi mans
[10:04:25] <kshishkov> and 1x8 should be enough to encode it in binary :)
[10:04:40] <mru> kshishkov: I should practise that
[10:04:50] <mru> it would annoy people for sure
[10:05:03] <kshishkov> you can start with Cyrillic
[10:05:09] <mru> 2x4 might be better
[10:05:17] <av500> 3x3?
[10:05:29] <av500> can be read ltr and ttb
[10:05:46] <mru> the blind use 2x3 fonts
[10:06:29] <av500> as a nice side effect, you can sell nice code also as a cross stitch pattern
[10:14:18] <pJok> hmm
[10:14:33] <pJok> my k800i has a 4 or 5px font
[10:14:43] <pJok> making it quite interesting to irc from with irssi
[10:28:40] <lu_zero> yawn
[10:29:14] <kshishkov> g'night
[10:29:36] <av500> night night kshishkov
[10:31:15] <cartman> what night
[10:35:41] <microchip_> m night shyamalan?
[10:35:43] <lu_zero> that night
[10:58:54] <cartman> "Program core engine was seriously improved to deliver even more...."
[10:59:00] <cartman> do that, "seriously improve"
[10:59:20] <kshishkov> add comments
[12:05:01] <KotH> mru: your secret gcc arm asm options blogentry... are the thumb-2 constant limits correct? -255 to +255 looks fishy
[12:05:06] <cartman> Does rtmpdump have a channel around?
[12:06:15] <kierank> no, only a mailing list
[12:06:21] <cartman> kierank: thanks
[12:06:24] <KotH> cartman: #mplayerdev and #ffmpeg-devel
[12:06:30] <cartman> their Makefiles are wrong for Mac
[12:06:33] <cartman> gonna fix it
[12:06:40] <KotH> cartman: hyc is the guy you want
[12:06:44] <av500> isnt that hyc toy?
[12:06:47] <cartman> KotH: cheers :)
[12:07:35] <mru> KotH: it's probably coded as sign+magnitude
[12:07:53] <KotH> mru: that'd be surprising
[12:08:18] <KotH> mru: 'cause it would mean that they'd have to do a sign+mag->2comp conversion
[12:08:38] <KotH> which is a lot more expensive than a sign extension in 2comp
[12:24:29] <cartman> patch done
[12:24:34] <cartman> now lets find hyc
[12:24:43] <cartman> hyc: ping
[13:06:11] <KotH> cartman: np: eski ordu marsi
[13:06:13] <KotH> cartman: :-)
[13:07:12] <mru> KotH: what are those constants for?
[13:08:07] <KotH> mru: which constants?
[13:08:27] <mru> the ones you were talking about earlier
[13:08:48] <KotH> ah.. i just read your blog a bit more carefully and stumbled over these values
[13:09:05] <KotH> they looked strange to my eyes used to twos complement
[13:09:24] <KotH> cartman: np: genc osman
[13:11:18] <mru> KotH: one of the T2 ldr encodings has an 8+sign-bit immediate offset
[13:11:36] <mru> it's probably meant to be used to generate offsets for that instruction
[13:12:48] <KotH> hmm...
[13:12:56] * KotH hasn't had a look at T2 yet
[13:13:04] <mru> don't bother
[13:13:07] <KotH> why not?
[13:13:08] <mru> it's a waste of time
[13:13:12] <KotH> why?
[13:13:46] <mru> small benefits, huge bugs
[13:15:23] <KotH> well, i will have to use an m3 soonish
[13:15:43] <mru> ah, T2 only
[13:15:45] <KotH> beside, it would be a good opportunity to learn arm asm in general
[13:15:54] <av500> KotH: which one?
[13:15:57] <mru> better pray you have a working compiler
[13:16:10] <KotH> mru: for gcc-values of working
[13:16:26] <av500> mru: I bet it comes with a nice Keil....
[13:16:28] <KotH> av500: at91sam3 for work, and probably an lm3s9xxxx for private use
[13:17:04] <KotH> av500: $customer is using keil... and complaining that half of the code doesnt compile correctly or throws warnings
[13:17:17] <KotH> even vor 7tdmi
[13:17:21] <KotH> s/vor/for/
[13:17:25] <av500> then run that half through gcc :)
[13:17:42] <KotH> but keil is much better[tm]!!!
[13:17:51] <av500> KotH: at the time of arm7 and arm9 we used the TI arm compiler
[13:18:16] <KotH> not to mention that said customer actually doesnt have the time to port our code from gcc to keil, but should be implementing the needed features
[13:18:31] <KotH> av500: never got my hands on that
[13:18:43] <av500> port? that bad?
[13:18:50] <KotH> yes, port
[13:19:23] <KotH> in approx 100kloc we've probably half a dozen places where we use inline asm... none of them complies (keil uses a different format)
[13:19:40] <KotH> all places where we use a little bit of "advanced" c, keil failes
[13:19:58] <av500> half a dozen places does not sound so much
[13:20:51] <KotH> we try to stay away from asm as far as possible
[13:21:33] <KotH> not many people here have experience with asm, and even fewer are able to use it efficiently
[13:21:47] <mru> keep it that way
[13:22:03] <mru> the fewer who know how to do it, the more I can charge...
[13:22:05] <KotH> what? the use of asm, or my coworkers understanding of asm?
[13:22:09] <KotH> lol
[13:22:24] <KotH> actually, you could teach me arm asm ;)
[13:22:41] <KotH> i'd pay richly... in swiss chocolate :-)
[13:23:16] <kshishkov> KotH: but can you afford $10k worth of swiss choco?
[13:23:42] <av500> kshishkov: seems that does buy much chocolate
[13:23:48] <av500> does *not*
[13:24:00] <KotH> kshishkov: no.. i'd just buy the cheap one and tell mru that it's been expensive :-)
[13:24:18] <av500> kshishkov: I saw the €2.5k amount :)
[13:24:25] <av500> KotH: :)
[13:24:35] <KotH> av500: 2.5k amount?
[13:27:58] <kshishkov> KotH: you're an engineer, just guesstimate how many boxes of Lindt it will take
[13:28:27] <mru> but you must guess professionally
[13:28:34] <av500> 54!
[13:28:41] <mru> factorial?
[13:28:47] <kshishkov> otherwise you're no engineer, like av500 here
[13:29:06] <av500> no idea why they graduated me
[13:29:17] <av500> guess it was the easiest to get rid of me
[13:29:47] <KotH> kshishkov: and which lindt? the prices per box vary iirc 1:30 or so
[13:30:04] <kshishkov> KotH: use the most expensive one, of course
[13:30:51] <KotH> kshishkov: 2860 boxes
[13:31:11] <KotH> kshishkov: but out of stock until oct/nov
[13:31:17] <twnqx> :O
[13:31:39] <kshishkov> KotH: ok, start saving for mru lessons
[13:31:41] <twnqx> KotH: send me some Kirschstengelei, my mother loves them, and i can't get them in decent quantity any more :X
[13:31:49] <twnqx> Kirschstengeli*
[13:31:49] <KotH> twnqx: lol
[13:32:00] * KotH will open a chocolate shop
[13:32:13] <KotH> twnqx: and what do i get in exchange? :)
[13:32:20] <twnqx> hm
[13:32:29] <twnqx> can i even offer anything but money? :X
[13:32:35] <av500> KotH: he can offer you BBB render time iirc
[13:32:41] <KotH> lol
[13:32:45] * KotH doenst need to render bbb
[13:32:47] <twnqx> uh, i did that...
[13:32:50] <KotH> twnqx: well.. i dont want money
[13:33:16] <twnqx> i'll... let you borrow my eagle pro license? :P
[13:33:33] <av500> the one from rapidshare?
[13:33:39] <twnqx> no, i bought it actually
[13:33:49] <mru> you can buy stuff on rapidshare?
[13:34:02] <KotH> twnqx: people get chocolate from me for: 1) being at lt/fosdem/froscon/... 2) sending me (good) chocolate 3) giving me access to documents i otherwise wouldnt get 4)....
[13:34:05] <av500> rapidshop
[13:34:11] <KotH> twnqx: eagle _pro_?
[13:34:16] <twnqx> yes
[13:34:21] <twnqx> the 1k$ license
[13:34:34] <av500> that defines the amount of shoggi then
[13:34:38] <twnqx> no autorouter, though
[13:34:54] <twnqx> i deemed it too useless to pay for it
[13:34:57] <KotH> twnqx: is that a kind of joke like boiled ice lollies?
[13:35:04] <twnqx> lol
[13:35:04] <av500> twnqx: I have a eagle spectra bridge somewhere
[13:35:27] <twnqx> KotH: it suits my needs for the time being :P
[13:35:47] <twnqx> i'll upgrade to better stuff should i require it
[13:35:47] <KotH> twnqx: hint: if a customer comes with an schematic or layout drawn with eagle, we charge twice the amount to redraw it properly than we'd do if we could start from scratch
[13:36:02] <twnqx> uh
[13:36:08] <twnqx> because you can't import it properly?
[13:36:18] <av500> KotH: so you like eagle then? :)
[13:36:20] <KotH> no, it's hard to get any production grade data out of eagle
[13:36:26] <twnqx> hm
[13:36:36] <twnqx> my pcb house so far did not have issues
[13:36:55] <av500> twnqx: DIL times are long over.. :)
[13:37:32] <twnqx> av500: 200µ copper traces
[13:37:35] <KotH> and nobody uses 1206 anymore, safe for huge ceramic capacitors
[13:38:03] <av500> twnqx: :) okok
[13:38:04] <twnqx> well, i use 0403 since that's around what i can do by hand :P
[13:38:15] <twnqx> err 0402
[13:38:24] <twnqx> i have some 01005 on my desk... cute :)
[13:39:08] <KotH> you've dust on your desk? ;-)
[13:39:21] <twnqx> yeah :p
[13:39:37] <twnqx> i have zero idea how to solder them by hand...
[13:39:59] <KotH> you cannot ^^'
[13:40:17] <KotH> 0201 already needs a very steady hand and special quipment
[13:40:33] <av500> you need small chinese hands
[13:40:36] <KotH> i actually dont even know how they handle 01005 on a machine... it's damn too small
[13:40:55] <av500> they just "sprinkle" them on the PCB
[13:41:01] <twnqx> heh
[13:41:18] <twnqx> put solder paste on, blow capacitors over pcb, hope they stick correctly? :>
[13:41:41] <av500> yeah
[13:42:00] <twnqx> well, i mostly use 0603..
[13:42:12] <KotH> that's called "self-assembly" and is all tha craze in research!
[13:42:27] <KotH> twnqx: who do you use for pcb fab?
[13:43:35] <twnqx> i think it was multipcb.de
[13:43:48] <av500> wasnt that beta layout?
[13:43:53] <twnqx> no
[13:43:59] <twnqx> beta layout looked like crap
[13:45:45] <av500> ah, right, beta is pcb-pool now
[13:45:49] <av500> mixed that up
[13:51:35] <KotH> twnqx: hmm... they seem to be quite cheap
[13:51:54] <KotH> twnqx: actually, quite a bit cheaper than pcbpool for euro card's
[13:52:07] <pJok> KotH, i have access to a semi-automatic smd placer :D
[13:52:08] <KotH> how's the quality?
[13:52:34] <KotH> pJok: i've access to an automatic smd solder machine: it's called Florian ;-)
[13:52:35] <twnqx> KotH: i used gold-plated contacts, and well, i think it was good enough
[13:52:48] <kshishkov> platinum is better
[13:52:49] <pJok> KotH. hehe
[13:52:50] <KotH> do you have pics of the pcb?
[13:52:58] <twnqx> i posted it before
[13:53:09] <av500> KotH: ours is called Madame Lam
[13:53:13] <twnqx> but i guess i can make new ones of pcb without components :P
[13:53:22] * pJok ponders on where to actually get that solderstop mask laquer along with something for the vias
[13:54:37] <KotH> pJok: electronic stores have solderstop laquer pens
[13:55:00] <pJok> pens aren't good enough, i want to be able to make something like the pcb's you can buy at those pcb companies
[13:55:15] <av500> ?
[13:55:33] <av500> make what?
[13:55:38] <pJok> the soldermask is easy, you just need some paint that will stick to plastics
[13:55:42] <pJok> av500, pcbs
[13:55:56] <twnqx> KotH: http://a.imageshack.us/img828/4654/dsc07141.jpg
[13:56:53] <KotH> twnqx: does your camera have a macro function?
[13:57:13] <KotH> twnqx: or do you have a microscope/binocular with a camera mount?
[13:57:20] <twnqx> that's not mine... i just grabbed that pic from my email :P
[13:57:35] <KotH> 'cause the interesting stuff is the one you only see when you look very closely :)
[13:57:37] <twnqx> i don't have a company registration, so i need to use other people to route my orders thtough
[13:57:53] <twnqx> and i'm in amsterdam, wuite a bit away from the PCBs :P
[13:57:57] * KotH would be shameless enough to use his company :)
[13:58:19] <twnqx> well
[13:58:26] <KotH> twnqx: you said that the fine wires are 0.2mm?
[13:58:36] <twnqx> what do you want to look at? friend just offered me a 600dpi scan :P
[13:58:39] <twnqx> yes
[13:59:47] <KotH> well, the interesting things are: 1) relative placement of coper to solderstop, 2) tolerances of via holes, 3) ireguarities of the trace edges
[14:00:00] <KotH> and why did you use gold plating?
[14:00:07] <av500> for da bling!
[14:00:30] <cartman> KotH: lol
[14:00:36] <cartman> weird choice of music
[14:00:46] <twnqx> a) i assumed i might have some of them around for longer => better shelf life
[14:00:49] <KotH> cartman: there is hardly anything better for working ;)
[14:00:57] <twnqx> b) easier to solder
[14:01:17] <twnqx> c) not THAT much more expensive
[14:01:51] <KotH> if a) is a problem, then they are not good at all. i dont think b) is true. c) is only an argument if you do it on company expense or have too much money
[14:02:16] <twnqx> copper oxidizes faster
[14:02:26] <KotH> er...
[14:02:35] <KotH> you always get a tin plating at least
[14:02:40] <av500> yep
[14:02:41] <twnqx> yeah well...
[14:02:54] <twnqx> wasn't there something with tin platings and flatness?
[14:02:56] <av500> but tin grows evil wiskers
[14:03:12] <twnqx> i was a bit afraid of those 0.4mm pitch ICs
[14:03:12] <av500> after a while they consume the PCB?
[14:03:29] <twnqx> lol
[14:03:35] <twnqx> that never happened to me so far :P
[14:04:02] <mru> evil whiskers? http://hardwarebug.org/wp-content/uploads/2010/02/dsc00508.jpg
[14:04:24] <av500> see, it comsumed the PCB...
[14:04:41] <kshishkov> and peed on another one
[14:04:59] <av500> gold plating could help there...
[14:05:14] <av500> lots of people have gold in the loo
[14:05:42] * kshishkov would like to have an emergency reserve of 1-2kg of gold in his loo too
[14:06:12] <av500> http://www.jmooneyham.com/solid-gold-toilet-for-the-rich.jpg
[14:06:30] <av500> http://3.bp.blogspot.com/_aH7Pw54-cjo/Rl1RWNlns5I/AAAAAAAAAMg/WI_orcTHzV4/s400/goldenes__klo.jpg
[14:08:46] <cartman> mru: nice kittens!
[14:12:32] <twnqx> KotH: http://a.imageshack.us/img84/9611/pcbcharlie.png
[14:45:16] <KotH> fucking suppliers!
[14:46:11] <KotH> i asked our assembly guy whether he can deliver the 300 pcb's we ordered for next week.. he said yes. now he called that his pcb supplier cannot deliver the pcbs until end of next week
[14:46:14] * av500 tends just to yell at them
[14:46:54] <KotH> anyone passing by schauffhausen tomorrow who could take a clue bat with him?
[14:48:17] <KotH> twnqx: did you use soldermaskd defined pads by purpose?
[14:50:12] <twnqx> yes
[14:50:19] <KotH> reason?
[14:50:33] <twnqx> iirc it was from their production data
[14:50:41] <KotH> interesting
[14:50:58] <twnqx> i think they say 100µ shift between the layers is possible
[14:51:07] <KotH> nobody uses soldermask defined pads with normal components
[14:51:17] <KotH> the only exception is BGA
[14:51:23] <KotH> and even there, it's frowned upon
[14:51:35] <av500> soldermask defined pads, whats that?
[14:51:47] <mru> pads defined by soldermask
[14:51:53] <av500> hmm
[14:52:13] <KotH> that's when your soldermask goes ontop of the copper of the pad and thus the shape of the solderable area is defined by soldermask
[14:52:20] <twnqx> hm
[14:52:44] <twnqx> i think it was because their minimum solder mask width was larger than what i needed
[14:52:46] <KotH> twnqx: other than that, the pcb looks not bad
[14:53:13] <twnqx> and remember the 0.4mm pitch ICs on the other side
[14:53:34] <KotH> with 0.4mm pitch, you dont have any soldermask inbetween the pads
[14:53:43] <twnqx> well...
[14:53:56] <twnqx> i wanted it still.
[14:54:15] <KotH> i'm actually suprised that they didnt remove the soldermask inbetween the pads of the QFP on top
[14:54:42] <twnqx> that's .5mm i think
[14:54:55] <twnqx> they can go as far as 0.2 for the mask
[14:54:59] <twnqx> minimum width
[14:55:03] <av500> KotH: y?
[14:56:43] <KotH> av500: if the soldermask lines get too thin, they break, and you end up with small sticks of soldermask pointing into the pads. thus most pcb suppliers just remove those
[14:57:00] <av500> but on .5 it should work
[14:57:04] <KotH> twnqx: we usually work w/o soldermaks between pads for anything from 0.63 down
[14:57:48] <KotH> twnqx: the assembly just has to be a bit more carefull with the solder paste
[14:57:58] <KotH> other than that, it's not a problem
[14:58:01] <twnqx> yeah, if you use paste...
[14:58:11] <twnqx> remember, i do regular hand soldering :P
[14:58:13] <KotH> if you do it by hand, it's even easier
[14:58:26] <twnqx> with 1l flux / pcb :P
[14:58:32] <KotH> lol
[14:58:32] <KotH> no
[14:58:49] <KotH> 0.3mm solder, binoc and a steady hand :)
[14:59:17] <av500> KotH: i prefer good mask and pretinned, it drops into the mask and you just heat it a bit :)
[15:00:23] <KotH> av500: works w/o mask too
[15:00:29] * KotH has done that...
[15:00:35] <av500> yeah
[15:00:36] <KotH> but it's not as reliable as normal hand soldering
[15:00:52] <KotH> 'cause the amount of solder/tin is very critical
[15:01:48] <av500> KotH: but mostly I let chinese worry about such stuff :)
[15:02:11] <mru> av500: lets you focus on worrying about the chinese
[15:02:17] <av500> yeah
[15:02:47] <av500> they steal all IP anyway, this way we make sure they make no errors stealing :)
[15:16:59] <twnqx> i don't have a steady hand.
[15:17:22] <mru> so you use butterflies?
[15:17:26] <av500> drink a steadifier...
[15:17:37] <mru> hmm, never seen those
[15:17:38] <twnqx> no, capillar effect soldering & liters of flux
[15:17:44] <mru> desteadifiers however...
[15:18:02] <mru> ah... the fumes...
[15:18:16] <av500> the fumes, they do something!
[15:18:35] <kshishkov> epoxy?
[15:18:58] <av500> kshishkov: was that popular in .au?
[15:19:02] <av500> .ua
[15:20:40] <kshishkov> could be
[15:21:04] <kshishkov> it will remove some jitter anyway
[15:21:23] <kshishkov> but the most advanced technology was called "assembling while sober"
[15:22:13] <kshishkov> it didn't really catch up though for some reason
[15:22:56] <av500> too experimental...
[16:43:58] <kierank> spaam: what have you done with ffmpeg-devel
[16:44:11] <spaam> nothing .
[16:52:44] <KotH> it would be a very sad day if he would be able to do anything
[16:52:55] <KotH> but it is nevertheless a good idea to blame spaam
[16:53:47] <spaam> pff
[16:54:03] <spaam> i did something good today.  that can lead to i get a job :P
[16:54:22] <KotH> but you didnt buy lobs any ice cream
[16:55:02] <spaam> but you didnt send him any chocolate from your great country :P
[16:55:15] <KotH> he didnt give me any ice cream either ;)
[16:56:19] <spaam> so i need to give you ice cream to get chocolate? :P
[17:00:21] <KotH> if you have some of lobs ice cream :)
[17:10:23] <ods15> 12:18:11 <KotH> kshishkov: you sound like ods15
[17:10:23] <ods15> 12:18:32 <KotH> "that glass of water was just 3 months salary"
[17:10:29] <ods15> ahhh, those days are long gone :)
[17:10:49] <mru> but it was fun!
[17:10:52] <ods15> today i make 3 times more in a single day than i did then in an entire month
[17:11:19] <ods15> the army? eh, sometimes fun :) the routine duties usually weren't
[17:11:28] <mru> no, mocking you was fun
[17:11:39] <ods15> yes, it was :P
[17:12:18] <ods15> i actually got a tiny scholarship from my university, and because i had nothing to do with it, i bought an htc desire
[17:15:02] <KotH> lol
[17:15:14] <KotH> you're getting dirty rich now? ;)
[17:15:30] <ods15> yup
[17:16:03] <ods15> i already saved about 240,000 shekels ($60,000 about?)
[17:16:03] <KotH> so, we'll see you soon at an event again?
[17:16:14] <KotH> ok, we will see you
[17:16:24] <KotH> you've no excuses now
[17:16:27] <KotH> none at all!
[17:16:32] <ods15> hehe
[17:19:42] <ods15> actually, my excuse is usually lack of time :) i'm working full time while studying
[17:20:32] <KotH> either your topic is too easy, or you arent really working and thus overpaid ;)
[17:20:58] <BBB> peloverde: ping
[17:20:58] <ods15> the latter
[17:21:04] <peloverde> pong
[17:21:15] <KotH> ods15: yappari!
[17:21:28] <ods15> it is much more acurate to say i'm *payed* for working full time, rather than actually working it
[17:21:31] <ods15> ?
[17:21:51] <KotH> lol
[17:21:56] * KotH would like that too
[17:22:16] <KotH> anways.. gtg..
[17:22:22] <KotH> have been here for too long already
[17:22:26] <KotH> have a nice evening
[17:22:27] <ods15> cya
[17:22:47] <cartman> see you KotH
[17:23:20] <CIA-92> ffmpeg: mstorsjo * r24696 /trunk/libavformat/rtpdec_xiph.c: rtpdec_xiph: Correct the bitmask for num_pkts
[17:24:31] <lu_zero> open64 seems working no
[17:24:32] <lu_zero> w
[17:35:00] <lu_zero> uhmm
[17:35:11] <lu_zero> point regarding yasm being hurtful
[17:35:19] <lu_zero> no ipa for you =P
[17:35:46] <lu_zero> (open64 doesn't mix ipa and non-ipa objects)
[17:38:29] <wbs> ipa?
[17:38:44] <mru> a type of beer
[17:41:32] <peloverde> do you ever look at CVEs and whatnot and wonder where they actually point out the bug?
[17:49:10] <lu_zero> inter process optimization
[17:49:34] <lu_zero> open64 4.2.3 doesn't work that well apparently
[17:49:40] <mru> procedure, no?
[17:49:45] <lu_zero> w/out ipa it cannot allocate
[17:49:48] <lu_zero> mru: righ
[17:49:53] <mru> and analysis, not optimisation
[17:49:54] <lu_zero> enough registers =P
[17:49:58] <mru> optimisation would presumably follow
[17:50:04] <lu_zero> yup
[17:50:11] <lu_zero> -Ofast triggers both
[17:50:23] <lu_zero> -ipa:  Perform interprocedural analysis and optimization
[17:58:40] <gmc> maybe this is a better place.. i've been trying to get ffserver running inside a freebsd jail but failed.. looking for the reason it appeared sin_family is not initialized in the sockaddr_in structure
[17:58:52] <gmc> initializing that fixed the problem of not being able to bind inside a freebsd jail
[18:00:02] <gmc> so, how to go about reporting this as a bug?? with a proposed patch??
[18:00:17] <wbs> send the patch to ffmpeg-devel
[18:00:30] <wbs> or write a bugreport at roundup.ffmpeg.org with the patch attached
[18:01:04] <gmc> you might want to mention that on http://ffmpeg.org/bugreports.html btw
[18:01:33] <wbs> that's for general bugreports without a patch
[18:01:55] <wbs> but perhaps it should be mentioned there
[18:02:04] <gmc> oh ok, i just couldn't figure out from the site where to take this issue
[18:02:34] <gmc> http://roundup.ffmpeg.org/issue1260 seems similar
[18:19:09] <peloverde> shit! I just blew away aacdec changes for aacx. I hate svn.
[18:19:19] <mru> use git
[18:19:46] <peloverde> I use git for aacx
[18:19:54] <peloverde> but I got sick of not being able to dcommit
[18:20:01] <peloverde> for ffmpeg
[18:20:12] <mru> make your own clone then
[18:20:30] <peloverde> I tired to and gave up hours later
[18:20:31] <mru> or clone mine
[18:20:43] <peloverde> The hack I had working on yours fell apart
[18:24:25] <peloverde> The git switchover can't get here soon enough :(
[18:48:14] <CIA-92> ffmpeg: mru * r24697 /trunk/ (10 files in 3 dirs):
[18:48:14] <CIA-92> ffmpeg: fate: move some groups of related tests to their own files
[18:48:14] <CIA-92> ffmpeg: This moves some groups of tests for single codecs to separate files,
[18:48:14] <CIA-92> ffmpeg: and adds shorthands for running all tests in a group.
[19:25:49] <BBB> wbs: working on your applehttp patch, give me a little, should be done in not too long
[19:34:58] <lu_zero> diff: tests/data/regression/acodec/alac: No such file or directory
[19:35:00] <lu_zero> uh?
[19:36:51] <lu_zero> somebody forgot alac?
[19:37:02] <mru> what are you doing?
[19:38:11] <lu_zero> make fate
[19:38:23] <lu_zero> (playing with open64)
[19:39:16] <lu_zero> http://pastie.org/1075817
[19:39:21] <mru> that means the test crashed and didn't create the output file
[19:39:28] <lu_zero> uhmm
[19:39:40] <lu_zero> how to make the whole thing more verbose?
[19:40:09] <mru> tests/data/regression/acodec/alac.err should contain any error messages from the test
[19:40:24] <lu_zero> nothing is there
[19:40:39] <lu_zero> dmesg helpfully says that indeed feng crashed
[19:40:41] <mru> err no
[19:40:48] <mru> it'll be somewhere else
[19:40:54] <mru> forgot where that file goes
[19:41:02] <mru> find tests -name alac.err
[19:41:09] <mru> or maybe acodec-alac.err
[19:41:26] <lu_zero> tests/data/fate/acodec-alac.err
[19:41:43] <mru> no, that's the wrong one
[19:41:46] <lu_zero> empty file
[19:42:00] <mru> look in tests/data/acodec/
[19:42:16] <lu_zero> tests/data/alac.acodec.err
[19:42:18] <lu_zero> found
[19:42:37] <mru> yes, the files should be organised better
[19:43:47] <lu_zero> uhmm
[19:44:02] <lu_zero> I could say that open64 seems not stable at all...
[19:44:16] <mru> hardly surprising
[19:44:19] <thresh> interested in make fate run on hpux/ia64?
[19:44:39] <thresh> oups, configure fails
[19:44:44] <mru> hehe
[19:44:52] <mru> no hpux support there
[19:45:15] <lu_zero> thresh: I was wondering how open64 is faring
[19:45:27] * lu_zero is looking for alternatives to gcc...
[19:45:40] <thresh> mru: http://dumpz.org/21454/
[19:45:59] <lu_zero> anything that links to libstdc++ seems doomed to fail
[19:46:24] <mru> thresh: get a real OS and try again
[19:46:49] <thresh> I can now rant about ffmpeg being selective :P
[19:47:23] <mru> this is the first time I've heard of anyone trying to build it on hpux
[19:47:45] <lu_zero> hpux is a real os on imaginary hardware
[19:47:50] <mru> try claiming it's linux or sunos and see what happens
[19:47:53] <mru> --target-os
[19:48:05] <mru> should work if you don't enable shared libs
[19:48:10] <thresh> lu_zero: hey, that itanium box is still alive :)
[19:49:53] <thresh> gmake fate fails, let me try with full build first
[19:52:34] <peloverde> wonderful, the results aren't the same and I'm not sure which was correct
[21:46:27] <skyphyr> hi all - I'm looking at writing an app to handle generating quicktimes for preview during post production
[21:46:36] <skyphyr> for this we'd be taking frame sequences and apply a LUT and then compressing those
[21:46:44] <skyphyr> the more I think about it the more sense it makes to add support to ffmpeg for the filetypes we need (I was thinking of using OpenImageIO for that) and a filter for the LUTs (using OpenColorIO for that)
[21:46:52] <skyphyr> does that sound like a good approach? is ffmpeg the right place to do this? (or should I be looking at something like transcode which leverages ffmpeg?)
[21:47:41] <Dark_Shikari> AVfilter would be where the LUTs would go
[21:48:00] <Dark_Shikari> what video formats do you need reverse-engineered that ffmpeg doesn't support?
[21:48:32] <skyphyr> it's for the source frames actually - openexr, dpx, sgi are the more likely ones
[21:48:52] <Dark_Shikari> that sounds like a job for imagemagick
[21:49:09] <skyphyr> ffmpeg supports imagemagick for conversion?
[21:49:24] <skyphyr> if we wanted to do an overlay then AVfilter would also be the appropriate place?
[21:49:47] <skyphyr> what bit-depth are the images in when AVfilter gets its hands on them?
[21:50:15] <Dark_Shikari> no, I mean that it might be better to use imagemagick for that kind of thing
[21:50:24] <Dark_Shikari> AVfilter can process in any bit depth its filters support
[21:50:34] <Dark_Shikari> pixfmt wise, ffmpeg supports 16-bit and 8-bit
[21:51:45] <skyphyr> okay - so if our source frames were in float then they'd need to be downsampled to 16/8 bit before it came through
[21:52:39] <Dark_Shikari> float o.0
[21:52:45] <Dark_Shikari> not only downsampled, you'd need it to be clipped
[21:52:46] <Dark_Shikari> er, clamped
[21:52:49] <Dark_Shikari> plus there's the problem of gamma
[21:52:50] <skyphyr> 32bits per channel per pixel
[21:53:08] <Dark_Shikari> ffmpeg doesn't really have infrastructure for gamma conversions and stuff
[21:53:28] <skyphyr> okay - so I really need to process everything to 8bit in sRGB space before it goes to ffmpeg
[21:54:08] <Dark_Shikari> what do you intend to encode to?
[21:54:33] <chiiph> hi... are there any plans to fix the -metadata key=value option to set flv's metadata?
[21:55:37] <skyphyr> quicktime, probably x264, 8-bit
[21:56:01] <Dark_Shikari> Then it'll need to be in YV12 eventually anyways
[21:56:22] <Dark_Shikari> so all you need to do is convert to something swscale supports
[21:56:25] <Dark_Shikari> then send it through swscale
[21:57:11] <skyphyr> thanks - I'll look through the docs on swscale
[21:57:33] <BBB> chiiph: I don't think flvenc supports metadata right now
[21:57:33] <Dark_Shikari> libavutil/pixfmt.h
[21:57:36] <skyphyr> makes sense to go straight from float RGB to YV12 then :)
[21:57:47] <skyphyr> ahhh - thanks Dark_Shikari :)
[21:58:41] <chiiph> BBB: any ideas on how can I set one key inside the metadata when I'm encoding and sending the encoded video to a streaming server?
[22:01:04] <mru> BBB, Dark_Shikari: you can now "make fate-vp8" to run all the vp8 tests
[22:01:36] <Dark_Shikari> mru: oh, nice.
[22:01:59] <mru> and the web page learned some new tricks
[22:02:28] <_av500_> rss feed?
[22:02:44] <mru> bah
[22:03:46] <_av500_> colors
[22:03:51] <_av500_> colours
[22:04:11] <mru> that's been there for a while
[22:04:26] <_av500_> little arrows
[22:04:31] <mru> try them
[22:04:36] <_av500_> and that focus highlight
[22:04:46] <mru> highlight is old
[22:05:03] <mru> only thing new today is arrows
[22:05:15] <_av500_> still cant sort by column headers
[22:06:17] <lu_zero> mru: did you add the table stuff I suggested?
[22:06:27] <mru> lu_zero: yes
[22:06:33] <lu_zero> great =)
[22:06:42] <mru> what difference does it make?
[22:06:55] <lu_zero> now you can add sorting with 3 lines =P
[22:06:59] <mru> no
[22:07:12] <lu_zero> clientside I can now ^^
[22:07:15] <mru> no
[22:07:20] <lu_zero> why not?
[22:07:36] <mru> because I just did something that breaks that
[22:07:48] <mru> unless you know some trick I don't
[22:08:06] <lu_zero> the failometer is the problem?
[22:08:09] <mru> no
[22:08:13] <mru> it's in the thead
[22:08:27] <mru> see those little arrows?
[22:08:46] <lu_zero> yup
[22:08:50] <lu_zero> nice =)
[22:08:51] <mru> click one
[22:09:09] <mru> that's done with hidden table rows
[22:09:17] <mru> so a dumb sort will break it
[22:10:32] <lu_zero> the first thing that come to my mind is put everything in a row
[22:10:42] <mru> how would you do that?
[22:10:50] <lu_zero> let me try
[22:11:02] <mru> a row is by definition one row
[22:11:19] <mru> short of wrapping each entry in a mini-table I see no solutin
[22:11:22] <mru> +o
[22:18:29] <Dark_Shikari> bah, I can never get gcc to use the high registers
[22:18:42] <Dark_Shikari> e.g. "cmp ah, dl" instead of "shr eax, 0x8, cmp al, dl"
[22:22:01] <lu_zero> btw
[22:22:01] <lu_zero> <tr style="display: none"><td></td></tr>
[22:22:05] <lu_zero> is strange
[22:22:23] <mru> better ideas?
[22:22:30] <Dark_Shikari> smells of wysiwig
[22:23:12] <mru> that's to keep the odd/even row count appear to only count visible rows
[22:23:34] <lu_zero> noticed now that I'm looking at the perl code ^^
[22:23:53] <mru> I guess I could just set different classes for the rows
[22:24:09] <mru> css seemed simpler at first
[22:24:41] <mru> Dark_Shikari: no wysiwyg here
[22:26:26] <kierank> mru: still don't do web ;)
[22:30:23] <BBB> chiiph: you can't
[22:30:36] <BBB> chiiph: it's not supported, you'll have to add code to flvenc to support that
[22:30:48] <chiiph> BBB: hmmm... ok... thank
[22:30:48] <chiiph> s
[22:31:15] <mru> kierank: I don't do web for hire
[22:31:47] <kierank> oic
[22:33:28] <lu_zero> uhmm using divs would be simpler if we want to keep exactly the same behaviour
[22:33:34] <CIA-92> ffmpeg: rbultje * r24698 /trunk/libavformat/mmst.c:
[22:33:34] <CIA-92> ffmpeg: Move send_media_packet_request() and clear_stream_buffers() up.
[22:33:34] <CIA-92> ffmpeg: Patch by Zhentan Feng <spyfeng gmail com>.
[22:33:41] <mru> lu_zero: using divs for what?
[22:34:28] <CIA-92> ffmpeg: rbultje * r24699 /trunk/libavformat/mmst.c:
[22:34:28] <CIA-92> ffmpeg: Remove is_playing variable.
[22:34:28] <CIA-92> ffmpeg: Patch by Zhentan Feng <spyfeng gmail com>.
[22:35:30] <Dark_Shikari> well, I'm going to fix the stupid cabac dependency
[22:35:33] <CIA-92> ffmpeg: rbultje * r24700 /trunk/libavformat/mmst.c:
[22:35:34] <CIA-92> ffmpeg: Move read_mms_packet() code to be inlined in the calling function.
[22:35:34] <CIA-92> ffmpeg: Patch by Zhentan Feng <spyfeng gmail com>.
[22:41:05] <janneg> wbs: ping
[22:41:24] <lu_zero> the tabular design is nice but is making that part _quite_ annoying ^^;
[22:41:55] <mru> this _is_ tabular data
[22:42:22] <mru> the 90s fashion of using tables for general layout was stupid
[22:42:22] <lu_zero> yup ^^
[22:42:31] <lu_zero> the opposite as well
[22:48:00] <lu_zero> the "few lines" option with this structure is gone =P
[22:49:02] <lu_zero> the best I could do to keep all the features is making the detail line being inserted/removed on click
[22:49:13] <lu_zero> (best -< no effort)
[22:57:42] <lu_zero> ok found the simplest possible way
[22:59:11] <lu_zero> repeating the td of the previous line and hidding them works as supposed
[22:59:18] <lu_zero> simple and ugly =P
[22:59:25] <mru> huh?
[22:59:46] <mru> that assumes a stable sort
[23:00:00] <mru> which most sorting functions are not
[23:00:32] <lu_zero> I should check if js says something
[23:00:36] <mru> anyway, I've done good old server-side sorting
[23:00:45] <lu_zero> better =P
[23:01:18] <mru> _av500_: did you hear that?
[23:01:22] <lu_zero> it would be broken by pagination
[23:03:10] <efriedma> mru: iirc, js sort is stable
[23:04:58] <CIA-92> ffmpeg: darkshikari * r24701 /trunk/libavcodec/ (vp56.h vp56rac.c vp5.c vp6.c Makefile vp8.c):
[23:04:58] <CIA-92> ffmpeg: VP5/6/8: eliminate CABAC dependency
[23:04:58] <CIA-92> ffmpeg: Create a custom table for VP5/6/8's renorm to avoid depending on H.264's.
[23:04:58] <CIA-92> ffmpeg: Saves one instruction in the arithmetic decoder as well.
[23:05:20] <mru> Dark_Shikari: hah, so it's not at all like h264 after all!
[23:05:37] <Dark_Shikari> mru: it's the same table
[23:05:39] <Dark_Shikari> with 1 subtracted from it
[23:05:53] <mru> </reddit>
[23:06:45] <lu_zero> efriedma: still if we add pagination the js one would need more code ^^;
[23:06:49] <_av500_> mru: i hear
[23:19:09] <efriedma> Dark_Shikari: it's also very similar to ff_log2_tab
[23:19:50] <mru> that means the evil mpeg has patented logarithms
[23:20:44] <lu_zero> mru: is there a way to coerce to int in perl?
[23:20:54] <mru> int $foo
[23:20:57] <Dark_Shikari> proposal:
[23:21:06] <Dark_Shikari> split h264dsp (the config name) to h264dsp and h264pred
[23:21:07] <lu_zero> if it isn't it gets 0?
[23:21:12] <Dark_Shikari> rv30, rv40, and vp8 get h264pred, but not h264dsp.
[23:21:31] <mru> ine
[23:21:32] <lu_zero> how many bytes spared?
[23:21:34] <mru> +f
[23:21:42] <Dark_Shikari> a lot
[23:21:57] <Dark_Shikari> I'm not quite sure what I have to do to make this work
[23:22:01] <Dark_Shikari> and whether it justifies any bumping of versions
[23:22:02] <lu_zero> no concerns regarding speed?
[23:22:15] <Kovensky> lu_zero: undef I think
[23:22:21] <lu_zero> that stuff should be all internal
[23:22:22] <Dark_Shikari> mru: http://pastebin.org/448107
[23:22:27] <Dark_Shikari> is there anything else that I'm missing?
[23:22:31] <Dark_Shikari> that I have to modify to do this
[23:22:43] <lu_zero> my $page = int param('page') || 0;
[23:22:48] <lu_zero> that should do then
[23:22:53] <Kovensky> nvm, retrns 0
[23:23:03] <Kovensky> int undef gives 0 too
[23:23:58] <mru> Dark_Shikari: look ok assuming those dependencies are in fact correct
[23:23:58] <Kovensky> lu_zero: put parens around the int, I'm not very sure about ||'s precedence but it might do param || 0
[23:24:13] <Dark_Shikari> mru: how does it get from "h264dsp" to "CONFIG_H264DSP"?
[23:24:22] <mru> magic
[23:24:41] <Dark_Shikari> ?
[23:24:50] <mru> there's a function somewhere that does that
[23:26:03] <Kovensky> lu_zero: "In other words, list operators tend to gobble up all arguments that follow, and then act like a simple TERM with regard to the preceding expression." (a function is treated as a list operator)
[23:26:16] <Kovensky> but I think that's related to commas, not to other stuff
[23:26:49] <mru> "int param || 0" actually gives same result as int(param) || 0
[23:27:10] <mru> and "int(param || 0)"
[23:27:17] <mru> precedence doesn't matter here
[23:27:32] <Kovensky> I was talking about precedence while I still thought int("garbage") returned undef :P
[23:27:43] <Kovensky> how do you check if something is int-like, regexp?
[23:28:32] <Kovensky> { int eq $_ } works, except if you have leading 0s
[23:28:54] <j0sh> lu_zero: how does rtsp return multiple packets from the same buffer without losing new incoming data?
[23:29:13] <lu_zero> http://pastie.org/1076261
[23:29:16] <lu_zero> j0sh: uh?
[23:29:48] <lu_zero> j0sh: point me to the code at hand
[23:29:51] <j0sh> read/fetch packet are called once when a new packet is ready (presumably meaning new data has been received from the stream)
[23:30:06] <j0sh> rtsp.c:1753, 1827
[23:30:10] <mru> lu_zero: but I don't want a paged view
[23:30:44] <lu_zero> we might need that once we surpass 100ish items
[23:30:50] <j0sh> lu_zero: (this is re: martin's patch to the xiph depacketizer, trying to figure out how it works at a low level)
[23:31:32] <mru> lu_zero: we're missing a handful of windows systems, that's all
[23:31:38] <mru> the list unlikely to grow much more
[23:31:47] <mru> for the history view it's another matter entirely
[23:31:59] <Dark_Shikari> mru: patch sent to list
[23:32:03] <Dark_Shikari> make sure I didn't miss anything in your arm realm
[23:34:47] <lu_zero> j0sh: rtpdec.c:383
[23:37:46] <j0sh> lu_zero: finalize_packet?
[23:38:10] <lu_zero> that function (we should remove the ts cruft btw)
[23:38:12] <lu_zero> does that
[23:39:09] <j0sh> but rtp_parse_packet only returns one packet
[23:39:36] <j0sh> and brings it back to rtsp
[23:39:45] <lu_zero> returning 1
[23:39:50] <lu_zero> more to do
[23:40:01] <lu_zero> rtsp calls back the rtp_parse_packet
[23:40:20] <j0sh> right
[23:40:41] <lu_zero> the buffer now is present, call to the parse_packet, return either 0 or 1
[23:40:42] <j0sh> but rtsp_fetch/read_packet (which calls rtsp_parse_packet) doesnt really do anything with the return value
[23:40:44] <lu_zero> and so on
[23:41:28] <j0sh> it just propagates the 1 up until rtsp_read_packet, which returns 0
[23:41:54] <lu_zero> looks like there is _something_ missing indeed then
[23:41:56] <j0sh> and presumably rtsp_read_packet isnt called again until there is new data coming in from the stream
[23:42:07] <Dark_Shikari> mru: requesting a comment or something on the ML
[23:42:15] <Dark_Shikari> (I don't want this to get bikeshedded or whatever, or left dead)
[23:42:58] <lu_zero> you are right
[23:43:42] <j0sh> we lose data?
[23:43:50] <lu_zero> wait
[23:44:25] <lu_zero> we shouldn't lose data
[23:44:47] <lu_zero> rtsp_read_packet <- ask for new data
[23:45:15] <j0sh> i cant figure out where the generic read_packet is called from
[23:45:33] <lu_zero> rtsp_fetch_packet <- outputs a packet and fetches it
[23:46:16] <j0sh> but rtsp_read_packet always outputs 0, short of an error... how does the caller know there are more packets waiting?
[23:47:42] <j0sh> right now it "works" because there is new incoming data in the stream, but i don't see where we handle both the packets in the old buffer, and the new data
[23:48:24] <lu_zero> av_read_frame_internal
[23:48:51] <lu_zero> j0sh: iirc ffmpeg doesn't work as you'd expect it working ^^;
[23:49:25] <Dark_Shikari> mru: thx
[23:53:04] <lu_zero> you call av_read_frame, it calls the read_packet, it calls the fetch, it calls the proto_read_packet and parses it or just parses it if something is leftover
[23:54:02] <lu_zero> I guess you were expecting that the were a loop driven by a poll/select somewhere
[23:54:05] <efriedma> Dark_Shikari: if you're looking into reducing the size of builds with stuff disabled, dsputil_mmx.c might be a good place to start; it's about a third of the size of a --disable-everything --enable-small build
[23:54:19] <Dark_Shikari> I already did that locally for gaikai
[23:54:25] <Dark_Shikari> the only reason I did this was because of bitching chromium idiots
[23:54:43] <Dark_Shikari> and by "did that" I mean "I just stripped out most of lavc"
[23:56:18] <lu_zero> Dark_Shikari: better having them bitching than having them keeping a local fork with those changes ^^;


More information about the FFmpeg-devel-irc mailing list