[FFmpeg-devel-irc] IRC log for 2011-03-03

irc at mansr.com irc at mansr.com
Fri Mar 4 01:00:05 CET 2011

[00:59:47] <Sean_McG> when using av_log(), how do you give the log entry a meaningful tag rather than [ NULL @ <addr> ] ?
[01:11:40] <astrange> pass a struct as the first argument whose first member is an AVClass *
[01:12:00] <mru> not worth the trouble for a quick debugging hack
[01:12:20] <Sean_McG> OK
[01:23:39] <Sean_McG> style question: true/false, TRUE/FALSE, or 0/1 ?
[01:24:01] <mru> !/nil
[01:24:10] <mru> if (cond)
[01:24:13] <mru> if (!cond)
[01:24:51] <mru> one day I'm going to sneak in #define TRUE 1, #define FALSE 2 somewhere
[01:25:00] <Sean_McG> LOL
[01:25:28] <mru> or #define TRUE 256 to fuck with people who typedef char bool
[01:41:03] <Sean_McG> when I reply to update a patch, which message ID should I reference... the last patch submission, or the latest reply?
[01:42:30] <mru> whatever makes sense
[01:43:22] <Sean_McG> OK
[01:44:32] <Sean_McG> sent
[02:44:11] <BBB> Yuvi: btw the ipad2 looks really nice. just mentioning, no reason why I'm pointing this out specifically to you, just, you know, saying... :-p
[02:47:57] <Sean_McG> hahah
[02:49:39] <kierank> BBB: omg now it's white
[02:50:17] <BBB> I know, it's got a new color, I gotta go buy one
[02:50:28] <BBB> well actually, that's unfair, I never had an ipad1 and pre-ordered the ipad2 a while ago
[03:02:58] <astrange> employee discount doesn't work on new products for a while
[03:05:29] <Jumpyshoes> BBB: how much is it gonna cost?
[03:05:49] <astrange> price is the same
[03:05:59] <Jumpyshoes> oh. is it just white?
[03:06:10] <astrange> http://store.apple.com/us/browse/home/shop_ipad/family/ipad/start?mco=OTY2ODA0NQ
[03:06:53] <astrange> http://store.apple.com/us/browse/home/specialdeals/ipad?mco=OTY2ODY4NQ
[03:07:04] <Jumpyshoes> that is expensive
[03:09:37] <astrange> refurbished is a good deal. there's really no reason to upgrade from the base model
[03:14:01] <Jumpyshoes> i don't buy apple products
[03:14:07] <Jumpyshoes> i was lucky enough to get a free itouch though
[03:59:33] <Jumpyshoes> what is nfc quantization?
[03:59:57] <mru> no f*cking clue
[04:01:11] <Jumpyshoes> it probably isn't important
[04:02:52] <kierank> google says noise feedback coding
[04:03:10] <Jumpyshoes> huh
[04:08:39] <Sean_McG> I'm with mru on that one ;)
[04:19:32] <Jumpyshoes> is there a way to make my cygwin configure nonretarded
[04:24:51] <Sean_McG> yeah, switch to *ix ;)
[04:25:49] <Shu> i should not have run make -j. i love it when cygwin freezes my entire box
[04:26:31] <kierank> haha
[04:26:50] <drv> i have to use cygwin at work, makes me die inside
[04:27:12] <Shu> i hate vim copy paste more than i hate cygwin
[04:36:26] <Jumpyshoes> AVERROR_PATCHWELCOME
[04:38:48] <Jumpyshoes> i didn't know this existed
[04:40:20] <Sean_McG> hahahahah
[04:47:42] <saintdev> peloverde: ping
[04:57:59] <Jumpyshoes> how do i just make ffmpeg? is it just "make ffmpeg"? it isn't work for me in cygwin
[05:00:10] <saintdev> me english parser isn't work either.
[05:00:44] <Jumpyshoes> how do i make ffmpeg and not ffplay, ffserver, etc?
[05:00:52] <kierank> --disable-ffplay --disable-ffserver
[05:00:56] <kierank> in ./configure
[05:01:10] <Jumpyshoes> oh great, another configure run <_<
[05:01:23] <saintdev> make ffmpeg works4meTM
[05:02:28] <Jumpyshoes> i bet cygwin is being retarded again
[05:21:45] <drv> you probably have to say make ffmpeg.exe
[05:26:52] <saintdev> drv: good point
[05:37:01] <peloverde> saintdev: pong
[05:44:08] <Jumpyshoes> can i get ffmpeg to surpress compiler warnings?
[05:46:39] <mru> close eyes
[05:48:16] <Jumpyshoes> well
[05:48:20] <Jumpyshoes> i do want to see compiler errors
[05:49:20] <ohsix> add -Werror
[05:49:29] <drv> haha
[05:52:56] <drv> wow, make -j (giantnumber) makes make significantly faster on cygwin
[05:53:18] <kierank> make isn't the problem on cygwin, it's configure
[05:53:29] <drv> yeah, configure is painful, but even make is super slow
[05:54:39] <ohsix> configure just runs lots of test programs, not good for cygwin
[05:55:38] <mru> configure runs the compiler, just like make does
[05:56:15] <ohsix> the tests are very short lived, and also ran themselves sometimes
[05:56:16] <elenril> mru: already awake or still awake
[05:56:34] <mru> still...
[05:57:00] <drv> the compiler runtime is almost negligible compared to the time spent sitting around waiting for process bringup and teardown
[05:57:29] <ohsix> yep
[05:57:40] <kierank> I guess I find configure painful on cygwin because you can't see anything going on. at least make looks like its doing something
[05:57:51] <ohsix> emulating the address space stuff on fork on windows is very very slow
[05:58:38] <j0sh> in swscale, are c and accelerated conversions supposed to return identical output?
[05:58:52] <j0sh> swscale-test with --disable-mmx returns totally different numbers
[05:59:17] <ohsix> "only with -bitexact" or something of that nature, i'd surmise
[05:59:55] <ohsix> but c & asm for filters should be the same i think (but don't know)
[05:59:56] <j0sh> hm
[06:03:33] <drv> hm, make -j12 is roughly 35 seconds vs make -j1 at 6m21s, which is pretty close to scaling linearly, just feels really slow
[06:04:23] <mru> on what hw?
[06:04:55] <drv> i7, 6 cores with ht
[06:05:07] * mru wishes he could use all the 900 cores on the supercomputer he's currently logged into
[06:05:38] <kierank> haha they're arguing about QCIF vs QSIF on jvt-experts now
[06:05:58] <mru> 240 vs 288
[06:06:34] <mru> I call them both epsilon
[06:07:51] <saintdev> kierank: in latin?
[06:08:00] <kierank> no, english today
[06:13:43] <saintdev> kierank: quad-QCIF?
[06:14:04] <kierank> lol
[06:14:41] <saintdev> QSIF-HD =P
[06:21:24] <kierank> mru: logged on to watson?
[06:22:18] <spaam> kierank: mru  is watson.
[06:22:37] <kierank> watson is powerpc iirc
[06:22:42] <kierank> mru must be an arm version
[06:23:07] <saintdev> POWER7
[06:24:39] <saintdev> This is a RAD video game codec.
[06:26:36] <kierank> What is codec?
[06:26:56] <Sean_McG> mmmm POWER7
[06:27:17] <saintdev> No, Watson.
[06:27:46] <KenJennings> What is Bink?
[06:28:20] <saintdev> lol
[08:16:22] <in3xes> ffmpeg
[08:16:35] <in3xes> oops! sorry
[08:17:57] <andoma> good morning internet!
[08:18:11] <Zor> why hello
[08:18:23] <thresh> moroning
[08:18:48] * av500 hands thresh some water
[08:25:07] <pJok> god morgon :)
[08:25:24] * pJok thought most of the drama was over on the ML
[08:28:33] * av500 lols at the QCIF drama on jvt-experts
[08:29:19] <thresh> av500: does it come with antisleeping pills?
[08:29:26] <av500> nah
[08:29:46] <av500> thresh: and its way to polite
[08:30:04] * thresh was asking about the water
[08:30:14] <av500> thresh: "little" warer
[08:30:17] <av500> thresh: "little" water
[08:30:23] <thresh> i'm puzzled
[08:30:41] <av500> I see that
[08:30:47] * av500 sends thresh back to bed
[08:41:50] <kshishkov> av500: link please
[08:46:34] <av500> kshishkov: to what? jvt?
[08:46:50] <kshishkov> yep
[08:46:57] <kshishkov> and to all their resources
[08:47:11] <av500> you want teh codez?
[08:50:47] <kshishkov> not sure
[08:51:51] <kshishkov> but I dislike the fact that most jvt-related places ask for login/password
[08:54:43] <av500> thats how one develops "open" standards, no? :)
[08:57:16] <thresh> http://lwn.net/SubscriberLink/430118/39152afef931d92e/
[08:58:58] <kshishkov> av500: I though the proper way is to buy proprietary stuff and release its sources as "open spec"
[09:01:15] <av500> thresh: that does not mention FFmpeg
[09:01:57] <kshishkov> or getopt.c
[09:03:11] <thresh> av500: I had an impression that bashing google is completely on-topic here
[09:03:46] <kshishkov> thresh: and here's a companion piece http://www.h-online.com/open/news/item/Controversy-surrounds-Red-Hat-s-obfuscated-source-code-release-1200554.html
[09:04:22] <thresh> yeah
[09:05:04] <av500> and?
[09:05:16] <av500> gplv4 can mandate .patch files
[09:05:38] <thresh> RHEL5 kernel comes with 2000 of those
[09:05:46] <thresh> no git whatsoever
[09:05:57] <av500> grep git GPL
[09:06:42] <thresh> ha ha you evil proprietor
[09:06:46] <av500> gplv5 can then mandate *all* GPL code to live in a git on stallmans gdium only
[09:07:26] <av500> oh, and its not online all the time, so print out patches and mail them....
[09:07:31] <kshishkov> av500: he has yeelong - gdium is not open enough for him
[09:07:49] <av500> yeehav then
[09:08:57] <av500> gplv6 can then mandata that only RMS can even write GPL code
[09:09:32] <kshishkov> av500: <offtopic> can you build new ArchOS player on http://www.loongson.cn/product_info.php?id=31 specially for him?</offtopic>
[09:09:58] <thresh> The source is configured: with a handwritten shell script [ +10 points of FAIL ]
[09:11:41] <kshishkov> The source is not configured at all [+100 points of FAIL]
[09:12:15] <av500> Your project only does releases as attachments in web forum posts [ +100 points of FAIL ]  --> ffforum.org
[09:13:58] <kshishkov> Linux kernel seems to gather lots of points on that test IIRC
[09:14:15] <av500> its a FAIL
[09:14:25] <kshishkov> GNU Hurd NG!
[09:14:32] <wbs> tanenbaum knew that 20 years ago, already
[09:14:33] <av500> it starts by being a fork of minix
[09:15:01] <av500> and linus was not involved in the parent project either
[09:16:00] <kierank> [08:52] kshishkov: but I dislike the fact that most jvt-related places ask for login/password --> not really
[09:16:03] <kierank> anyone can register
[09:16:12] <ohsix> fork of minix?
[09:16:17] <av500> so why ask for a password then?
[09:16:34] <av500> ohsix: it uses 3 letters
[09:16:45] <av500> out of 4
[09:16:54] <av500> clearly a fork
[09:17:01] <kierank> av500: to stop spammers and "plz send me the codes types"
[09:17:07] <ohsix> ah, good one; awesome to start flamin' too
[09:17:08] <kierank> doesn't work 100% i must add
[09:17:19] <kshishkov> doesn't work at all I think
[09:17:35] <kshishkov> because Chinese/Indians are hard-working
[09:17:37] <av500> kierank: signing up to a ml is one thing
[09:17:49] <av500> reading the archives behind a pwd is another
[09:19:12] <thresh> so ffmpeg gets 10 points of FAIL, and vlc only 5 points of FAIL
[09:19:15] <kshishkov> av500: that's to prevent spammers reading if somebody sent the codez so they can ask it immediately after registration
[09:19:33] <kshishkov> thresh: autotools count for +666 point of FAIL
[09:19:44] <thresh> autotools <3
[09:19:55] <ohsix> also a good place to start a flame
[09:41:20] <cartman> moin
[09:42:39] <av500> cartman: packed?
[09:43:27] <cartman> av500: packed for a small holiday, flying next Wednesday! :)
[09:43:45] <av500> dont come via frankfurt, its not safe
[09:44:19] <kshishkov> only for brain
[09:44:45] <kshishkov> and fortunately München airport should be ideal for him
[09:44:49] <cartman> av500: coming directly to Nürnberg
[09:45:35] <kshishkov> av500: BTW, what do _you_ have against the airport?
[09:45:48] <av500> kshishkov: http://www.dawn.com/2011/03/03/two-us-airmen-killed-in-frankfurt-airport-shooting.html
[09:46:53] <av500> kshishkov: I have nothing against it
[09:46:56] <kshishkov> av500: Kosovo, US, Frankfurt - a dangerous combination indeed
[09:46:58] <cartman> yeah :/
[09:47:15] <av500> kshishkov: now add KKP people
[09:47:21] <cartman> lol
[09:47:57] <kshishkov> av500: what is KKP? Some Chinese newspaper?
[09:48:54] <pJok> god morgon kshishkov
[09:49:02] <kshishkov> goda morgnar, pJok
[09:49:13] <av500> kshishkov: no, cheap trick to fool cartman's controlling officers
[09:49:39] <cartman> hahah :)
[09:50:05] <kshishkov> cartman: well, we know you're all for free Kurdistan ;)
[09:50:13] <av500> kshishkov: damn yoi
[09:50:15] <av500> u
[09:50:31] <av500> suse cannot afford to lose another employee
[09:50:38] <av500> they are so few already
[09:50:46] <kshishkov> good riddance!
[09:52:38] <cartman> bah
[09:52:50] <cartman> wait one week and then you can joke all day long :P
[09:53:29] <av500> then it wont be fun
[09:53:51] <cartman> true :P
[09:53:51] <av500> until we learn your BND trigger words
[09:54:56] <kshishkov> cartman: anyway, you'll be in Bavaria and we in Germany
[09:55:13] <cartman> I am OK with that
[09:55:21] <cartman> as long as visa says Germany
[09:55:26] <kshishkov> :)
[09:56:01] <kshishkov> don't forget to try Bayrische traditionell Weißdöner and Leberköfte
[09:56:27] <cartman> I'll make sure :P
[09:56:46] <thresh> uhmm, bavarian dirndls
[09:57:05] <cartman> http://placekitten.com/
[09:59:19] <kshishkov> thresh: сарафаны с кокошниками лучше?
[10:00:02] <thresh> kshishkov: "uhmm" означало http://4.bp.blogspot.com/_dH3Lwg9Gc1Y/SXfkSS2puPI/AAAAAAAAAPc/9528CDmviHg/s320/drooling_homer-712749.gif
[10:04:23] <av500> BSAC, do I ever need that?
[10:05:46] <kierank> av500: some korean or japanese thing uses it iirc
[10:06:56] <kshishkov> kierank: then they are ven crazier than anybody thought
[10:07:08] <peloverde> korean broadcast tv uses bsac
[10:07:39] <peloverde> t-dmb
[10:08:14] <av500> yes
[10:08:19] <av500> besides dmb
[10:10:46] <kshishkov> that's very dmb IMO
[10:10:56] <thresh> It also counts for substantial FAIL if "all your web site has is a picture of a marijuana leaf," as he said one small-scale open source project does.
[10:11:10] <kshishkov> what convenient features does BSAC have?
[10:11:47] <kierank> kshishkov: the feature that it is in 14496-3
[10:13:39] * kshishkov thinks it's very obvious that there are a lot of things in 14496-3
[10:14:08] <kshishkov> why not TwinVQ then?
[10:14:43] <kshishkov> or somebody just wanted to exercise arith coding patents belonging to them?
[10:14:49] <kierank> of course
[10:15:24] <peloverde> fgs bullshit
[10:16:09] <kshishkov> fgs?
[10:17:11] <peloverde> fine grained scalability
[10:18:41] <kshishkov> then why it's allowed to have SBR-in-BSAC ?
[10:21:20] <peloverde> bsac still has arithmetic coding
[10:22:39] <peloverde> and doesn't bsac+sbr use "Scalable sbr"?
[10:24:19] * kshishkov wonders how well "scalable" and "arithmetic coding" go along
[10:25:22] <kierank> kshishkov: anything can be scalable when you're mpeg
[10:26:04] <kierank> like hevc scalable
[10:26:44] <kshishkov> kierank: your phrase sounded like one of the songs from the South Park movie
[10:27:06] <pJok> anything is scalable if you use just add enough math to it
[10:27:48] <kshishkov> pJok: if you add enough cores
[10:29:32] <kshishkov> kierank: just look at the last lines - http://www.metrolyrics.com/im-super-lyrics-south-park.html
[10:30:40] * kshishkov lols hysterically at "byte-swapped IMA ADPCM data" on ML
[10:32:49] * av500 suggests a parser
[10:33:16] <av500> or libavswap
[10:34:18] <kshishkov> thanks, he decided to use parser already
[10:34:41] <kshishkov> I'd propose bitstream filter instead
[10:34:45] <kierank> lol libavswap
[10:35:20] <av500> pmi, but why was wrong about just swapping the damn bytes in the RM demuxer?
[10:35:26] <av500> why->what
[10:36:33] <kshishkov> it's a hack
[10:38:30] <av500> kshishkov: so is the de-interleaving of cook also a hack and should be done in an outside parser?
[10:38:30] * kshishkov found that RM specs has only one instance of "DNET" - the word "idnetify"
[10:38:37] <kshishkov> av500: nope
[10:38:48] <kshishkov> actually RM specifies a number of interleavers
[10:38:50] <av500> kshishkov: dnet is a 2byte interleaving :)
[10:38:54] <av500> solved
[10:38:57] <av500> next
[10:39:07] <kshishkov> av500: nope
[10:39:30] <kshishkov> files with it have "Int0" interleaver - i.e. no interleaving was done
[10:40:54] <av500> jaja
[10:41:25] <av500> just go ahead and add more memcpy
[10:51:28] <kierank> plz send me codez for memcpy
[10:52:03] * cartman hands kierank a 0day memmove
[10:52:18] <kshishkov> cartman: memmove is more robust
[10:52:27] <av500> kierank: here: http://www.gtk.org/
[10:52:45] <av500> ah, no that was teh codez for strcmp
[10:52:59] <kierank> av500: same thing, no?
[10:53:02] <cartman> kshishkov: no need for overlap check
[11:03:30] <kshishkov> mate!
[11:06:44] <pross-au> ksh
[11:07:45] <kierank> kshishkov: you should go and visit pross-au
[11:08:06] <pross-au> Bring ya swag
[11:09:19] <kshishkov> pross-au: http://zod.sourceforge.net/
[11:09:27] <pross-au> Cmdr Zod?
[11:09:52] <pross-au> Hey cool
[11:09:52] * pJok wonders how a codec designed by ffmpeg commitee would look like
[11:09:52] <kshishkov> just look at it
[11:10:05] <pJok> rather, format... audio+video+container
[11:10:13] <kshishkov> pross-au: so guess your work would be appreciated, mate
[11:10:48] <kshishkov> pJok: never completed due to bikesheds
[11:11:34] <pJok> kshishkov, possibly...
[11:11:40] <pross-au> consensus was obtained on the container
[11:11:49] <pJok> kshishkov, isn't ffv1 designed by ffmpeg committee?
[11:12:34] <Dark_Shikari> you mean michael
[11:12:47] <kshishkov> pJok: nope, ffv1/snow and maybe nut were designed by one man
[11:12:53] <pJok> ah
[11:13:10] <pJok> Dark_Shikari, tomato potato
[11:22:04] <iive> kshishkov: this is a lie
[11:22:49] <iive> at least the nut bit.
[11:23:01] <kshishkov> that's why I said "and maybe NUT"
[11:23:10] <kshishkov> I know Oded worked on it
[11:23:14] <iive> then don't say it.
[11:25:52] <pross-au> somebody should really survey usage of the many formats available.. would make for an interesting read
[11:27:59] <iive> pross-au: avi,mkv,mp4,ogm in that order.
[11:29:36] <pross-au> what? no .bik!
[11:29:42] <av500> .rmvb
[11:29:48] <av500> all of china uses that
[11:30:03] <cartman> indeed
[11:30:05] <av500> and .wmv for all the pron
[11:30:10] <iive> not really, they are still at vcd :P
[11:30:13] <cartman> av500: I wonder why though (rmvb)
[11:30:16] <pross-au> could be a fun survey then
[11:30:20] <av500> cartman: nobody know
[11:30:22] <av500> cartman: nobody knows
[11:30:27] <av500> even real does not know
[11:30:31] <av500> we asked them
[11:30:33] <cartman> lol
[11:30:48] <cartman> we have 1080p rmvb samples from China
[11:30:50] <cartman> amazing
[11:30:59] <av500> cartman: theory is that at some point in time, real producer "worked" for the chinese
[11:31:04] <av500> and these people just stuck to it
[11:31:07] <cartman> :D
[11:31:19] <av500> or it was the easiest to cracj
[11:31:20] <av500> k
[11:39:13] <pross-au> and avi is popular because, its legacy, good enuff?
[11:39:25] <kshishkov> yep
[11:39:36] <kshishkov> be thankful WTV is not that common
[11:39:52] <pross-au> Windows has had built in support for WMV for over a decade now
[11:40:38] <pross-au> truth be told, i started thinking about a wtv muxer
[11:44:55] <pJok> pross-au, just to mess with wtv specs?
[11:46:26] <pross-au> to help build a better demuxer
[11:46:44] <pross-au> because theres a lot of unknowns in there
[11:48:16] <pross-au> it seems any modules/filters int the wtv streaming graph can write chunks, so theres a lot of proprietary chunks in the wtv file
[11:51:51] <kshishkov> so it's completely the same in design as OLE archives (aka M$ Office document formats)
[11:52:15] <pJok> kshishkov, not that odf is much better
[11:53:04] <cartman> [applehttp @ 0xc19180]Receiving 1 variant streams
[11:53:07] <cartman> awesome :D
[11:53:19] <pross-au> kshishkov: precisely
[11:53:26] <spaam> cartman: wait until you have 2 variants..
[11:53:39] <av500> or 42
[11:53:41] <cartman> was 8 variants
[11:53:45] <cartman> it autoselected
[11:53:46] <cartman> :D
[11:54:09] <cartman> now I can watch our turtleneck Steve showing iPad2
[11:54:34] <Compn> variants?
[11:54:37] <Compn> sounds like mbr
[11:55:38] <Compn> [06:45] <av500> we asked them
[11:55:40] <pJok> cartman, good ol' appleSteve
[11:55:40] <Compn> lol!
[11:55:42] <Tjoppen> no negative feedback on my wav patches at all? this must be a first :)
[11:55:59] <mru> Tjoppen: you got a wtf from michaelni
[11:56:09] <cartman> pJok: indeed
[11:56:17] <Compn> av500 : next you'll have to ask china , but i guess its 'what everyone else is using' like divx/h264 in the rest of the world
[11:56:44] <Tjoppen> I don't see why. I rearranged the some code that was already there
[11:57:59] <av500> Compn: yes
[11:58:09] <av500> Compn: they even did a survey back in the days
[11:58:28] <av500> Compn: and found that .rmvb was only used for P2P content :)
[11:59:51] <merbzt> Tjoppen: indentation I think
[12:01:10] * Compn wonders how many hardware rmvb players there are in china now
[12:01:17] <Compn> using ffmpeg code
[12:02:33] <spaam> cartman: 2
[12:04:55] <Tjoppen> then he must have missed that it's rearranged code
[12:08:58] <iive> Tjoppen: i think it have more to do with the possibility of seeking to int64_max position.
[12:09:26] <iive> or rather offset.
[12:13:02] <Tjoppen> hm
[12:13:22] <Tjoppen> right, might not want to set next_tag_ofs to INT64_MAX
[12:24:45] <mru> #ifdef Is_True_On
[12:25:08] <Tjoppen> enum { TRUE, FALSE, FILE_NOT_FOUND };
[12:59:08] <thresh> kshishkov: http://lenta.ru/news/2011/03/03/hmao/
[13:00:45] <kshishkov> thresh: ну и страна, однако
[13:02:44] <JEEB> Великая страна Россия
[13:06:15] <kshishkov> JEEB: so would you prefer to live in Grand Duchy of Finland or Österbotten (if current country is not available)?
[14:20:50] <in3xes> What is the  correct place for title, author, year etc.?
[14:24:19] <twnqx> isn't metadata just built as a list of attribute/value pairs?
[14:25:05] <kierank> ask elenril
[14:27:09] <JEEBsv> kshishkov: I'm not sure. Might choose Österbotten.
[14:27:13] <JEEBsv> My name is fully Swedish anyways
[14:27:14] <JEEBsv> lol
[14:37:33] <in3xes> elenril: ping
[14:44:52] <J_Darnley> What do you mean by "correct place"?
[14:52:35] <in3xes> J_Darnley: Fields in AVFormatContext are deprecated. So where should I put them?
[14:52:56] <J_Darnley> The AVMetadata struct
[14:53:10] <J_Darnley> Using the supplied functions
[14:53:21] <elenril> av_metadata_set2
[14:53:35] <J_Darnley> like that ^
[14:53:48] <in3xes> Okay
[14:54:30] <kshishkov> elenril: today I ate lie from a can
[14:59:37] <elenril> http://tvtropes.org/pmwiki/pmwiki.php/Main/BlatantLies
[15:00:00] <kshishkov> elenril: nope, different lie - the one featured in Portal
[15:00:33] <elenril> there really was a cake
[15:00:37] <elenril> and it was delicious too
[15:01:12] <kshishkov> yes, since I happen to live in a region where that cake originated from, I tried it
[15:03:58] <elenril> btw
[15:04:06] <elenril> anyone wants to mentor playlists?
[15:14:55] <in3xes> Isn't that an abandoned project?
[15:15:43] <elenril> yeah, it failed in the last gsoc
[15:16:23] <in3xes> btw, I am a student :)
[15:16:38] * elenril is a student too
[15:20:20] <BBB> elenril: you want to do playlists?
[15:20:24] <BBB> elenril: that's a massive task
[15:20:29] <BBB> elenril: I thought you were busy
[15:20:41] <BBB> in3xes: want to participate in gsoc?
[15:21:03] <elenril> BBB: can't be THAT bad
[15:21:16] <BBB> I've been told it eats your soul
[15:21:23] <elenril> maybe i could recycle something from the previous attempt
[15:43:12] <BBB> elenril: if possible
[15:43:17] <BBB> elenril: would you mind going through that code?
[15:43:41] <BBB> elenril: my impression was that the student sort of went off by himself, didn't discuss and so some code might be useful but the concept at large isn't very useful
[15:43:49] <BBB> i.e. first plan how to integrate it, then code
[15:43:52] <BBB> not the other way around
[15:51:08] <in3xes> BBB: yes
[15:53:11] <BBB> in3xes: what project (or type of work) are you interested in?
[15:53:24] <BBB> elenril: btw I'm all for you doing playlists, you'll have a much higher chance of success than anyone else
[15:53:40] <BBB> I'm just affraid it's like the kernel block layer: waterloo for everyone
[15:54:03] <elenril> what's wrong with kernel block layer?
[15:54:10] <elenril> and who'd mentor it? you?
[15:54:18] <Kovensky> BBB: stalingrad > waterloo
[15:54:56] * av500 proposes a neon audioconvert and downmix
[15:55:03] <av500> mru to mentor it
[15:55:06] <BBB> elenril: dunno... we'll have to find someone
[15:55:40] <mru> Kovensky: hardly, stalingrad is an obscure paris metro station, waterloo is one of the main rail hubs in london
[15:56:39] <Kovensky> ofc london would name one of their main rail hubs after one of the most famous defeats for france :D
[15:56:44] <kierank> yup
[15:56:56] <kierank> the rail hub that had trains to france too
[15:57:35] <mru> and it's right next to trafalgar square
[15:57:52] <mru> baptiste mumbled something when he visited...
[15:58:29] <elenril> maybe wbs...
[15:58:54] <kshishkov> Kovensky: well, all Russian/Ukrainian railroad stations are named after one in London
[15:58:56] <in3xes> BBB: Any suggestion?
[15:59:07] <kierank> mru: nope, waterloo's on the other side of the thames
[15:59:37] <mru> close proximity
[15:59:44] <mru> it's <10m walk
[15:59:53] <kierank> yeah i suppose
[16:00:37] <mru> the various rail/tube stations in that area are weird
[16:03:33] <kierank> a lot of them are right next to each other because in the 1800s competing rail companies built them there
[16:06:51] <Kovensky> http://www.cracked.com/article_18550_5-true-war-stories-that-put-every-action-movie-to-shame.html <-- #5 on why stalingrad > waterloo
[16:09:35] <j-b> hello people!
[16:09:56] <lu_zero> hi
[16:10:13] <mru> hello french
[16:11:42] <av500> j-b: booze on monday?
[16:12:45] <j-b> av500: NO!
[16:12:56] <j-b> av500: I take a true week of holidays :D
[16:13:01] <av500> pah
[16:13:06] <j-b> like no Internetz
[16:13:07] <av500> lazy french
[16:13:26] <j-b> French, yeah... Lazy, no
[16:18:26] <av500> why is audioconvert.h not exported?
[16:18:36] <av500> what am i missing?
[16:19:56] <kierank> in libavutil?
[16:23:58] <av500> kierank: ffplay.c uses av_audio_convert
[16:24:27] <av500> I want to use it too
[16:24:36] <av500> but I fail to find it in installed headers
[16:24:39] * av500 feels stupid
[16:25:04] <kierank> it is in mine
[16:25:08] <kierank> oh wait
[16:25:09] <kierank> no it isn't
[16:25:45] * av500 is trying a codec that returns goddam float values
[16:51:58] <av500> kierank: ok, I replaced all of audioconvert with one line I stole from it...
[16:51:59] <thresh> paris has a stalingrad metro station? most bizarre
[16:53:06] * kierank adds av500 to hall of shame
[16:53:24] * av500 laughs evil laugh
[17:00:01] <mru> we should have a hall of flame
[17:07:54] <BBB> elenril: as for the cafdec patch, I'd recommend to leave that for now, I don't think it really helps all that much
[17:09:05] <elenril> get_strz has to be replaced anyway
[17:09:53] <elenril> and since the size of the chunk is already known (and other parts of the code already rely on it being correct), we might as well use it
[17:15:00] <BBB> the whole file appears to not deal with, say, the case where there's a few padding bytes after a chunk
[17:15:13] <BBB> so I wouldn't say that the size is used at all
[17:21:27] <elenril> yeah, i noticed that too
[17:21:32] <elenril> but it's used in other places
[17:22:59] <elenril> ok, i'll remove it from the patch
[17:23:17] <elenril> not like i care about this demuxer
[17:26:27] <BBB> elenril: that's what I thought also, like, I don't even know what caf is :-p
[17:26:40] <BBB> elenril: "thanks for not caring" sounds weird though
[18:33:23] <elenril> sure is netsplitty
[18:33:36] <lu_zero> a lot
[18:33:38] <elenril> do we turn url_ftell into a macro as well?
[18:43:52] <elenril> ffmpeg can't dump attachments, right
[18:44:21] <elenril> lu_zero: weren't you working on this?
[18:46:15] <lu_zero> elenril: posted a patch
[18:47:19] * elenril searches for it
[18:54:21] <elenril> hmmm....fate still fails for me
[18:54:26] * elenril wonders 
[19:05:14] <lu_zero> elenril: found it?
[19:07:12] <elenril> yeah, will try now
[19:08:31] <lu_zero> it should be able to streamcopy data streams
[19:08:42] <lu_zero> well it is for nut -> anything
[19:09:31] <elenril> what i want is to dump an attachment into file
[20:28:25] <BBB> elenril: thanks
[21:12:10] <merbanan> where is the gitorious clone ?
[21:15:59] <lu_zero> http://gitorious.org/ffmpeg/ffmpeg I think
[21:16:19] <merbanan> can you send a patch for the homepage
[21:21:21] <Compn> huh ?
[21:21:36] <Compn> ffmpeg.org homepage you mean ?
[21:22:08] <merbanan> y
[21:23:03] <Compn> merbanan : svn://svn.mplayerhq.hu/ffmpeg.org
[21:23:03] <Compn> iirc
[21:23:22] <Compn> E:\mplayer-testclips\ffmpeg.org>svn info
[21:23:23] <Compn> Path: .
[21:23:23] <Compn> URL: svn://svn.mplayerhq.hu/ffmpeg.org
[21:23:36] <Compn> yep thats it
[22:33:29] <Sean_McG> hi folks
[22:35:24] <gnafu> Howdy, Sean_McG!
[22:36:24] <Sean_McG> how goes?
[22:52:34] <Jumpyshoes> how do you find the MSB in a int? i.e. ceil of log2(x)
[22:58:54] <mru> Jumpyshoes: count leading zeros
[22:59:09] <mru> in ffmpeg you can do av_log2()
[22:59:16] <Jumpyshoes> oh, okay
[22:59:19] <mru> it does whatever is best on your cpu
[23:39:05] * Kovensky hurfs here too
[23:39:07] <Kovensky> 20:36.37 Kovensky hurfs @ the vector math professor
[23:39:07] <Kovensky> 20:37.02 Kovensky: he was doing pretty well... until he said that a 3D vector with the constraint "|y| = z" could point to any octant :|
[23:39:10] <Kovensky> 20:37.31 Kovensky: and he insisted on it despite my objections
[23:39:14] <Kovensky> 20:37.48 Kovensky: he compared "|y| = z" to "z = ±y", despite it making no sense
[23:39:17] <Kovensky> 20:38.09 Kovensky: and found hard to understand when I said that "|y| = z" implies that z is positive
[23:40:10] <mru> actually, it's non-negative
[23:40:19] <mru> positive means strictly >0
[23:40:26] <mru> but you're right
[23:41:53] <Kovensky> hopefully that's the extent of the confusion :/
[23:42:13] <Kovensky> in b4 he asks it on a test and expects the "points anywhere" answer
[23:42:53] <Jumpyshoes> Kovensky: vector math as in multivariate calculus?
[23:43:41] <Kovensky> idk yet
[23:43:45] <Kovensky> this is the 3rd day of uni =p
[23:43:58] <Jumpyshoes> well what was in the course description?
[23:44:02] <Kovensky> (calculus is strictly an university level topic in brazil)
[23:44:19] <Kovensky> I don't have access to those yet, only when I get my student number
[23:44:41] <Jumpyshoes> how do you know what classes you're taking then ._.
[23:44:59] <Kovensky> 1st semester students automatically are assigned to the 1st semester modules, and they give you a table with the modules/hours
[23:45:33] <Jumpyshoes> i see
[23:46:18] <Kovensky> I should receive mine sometime next week if there are no delays
[23:47:06] <Jumpyshoes> so are you in the US equivalent of high school?
[23:47:43] <Kovensky> equivalent of university, actually
[23:47:46] <mru> vector calculus is calculus with vector-valued functions
[23:47:59] <Kovensky> TJ is the weird one, based on what I hear about american schools ._.
[23:48:29] <Jumpyshoes> i see
[23:48:38] <Kovensky> (interestingly, in both portuguese and swedish, education levels are "lower" than in english)
[23:48:45] <mru> so instead of y = f(x) you have (t,u,v) = f(x,y,z)
[23:48:49] <mru> if in 3 dimensions
[23:48:57] <mru> you can have any number of dimensions
[23:49:02] <Jumpyshoes> you should cover vector calculus in multivariate calculus
[23:49:30] <Kovensky> such as "high school" being 'ensino médio' ("middle school"), university being the "high school"
[23:49:57] <Kovensky> idk if this inversion exists in swedish, but undergraduation in english is called graduation in portuguese, and english's graduation is post-graduation in pt
[23:50:21] <Jumpyshoes> i've always heard it as elementary school --> middle school --> high school --> college/university --> grad school --> PhD
[23:50:37] <Kovensky> talking about post-sth, reminds me of brazilian discordians translating aftermath to the equivalent of "post-mathemathics"
[23:50:57] <Kovensky> (aftermath being the last month of the discordian year)
[23:51:11] <Jumpyshoes> what in the world is aftermath
[23:51:24] <mru> in swedish "högskola" (literally "high school") is a not-quite-university
[23:51:43] <mru> they teach at the same level but aren't authorised to give out certain degrees
[23:52:54] <mru> aftermath is what remains after a disaster, such as a maths degree
[23:53:29] <Jumpyshoes> lol
[23:53:49] <BBB> yiihaa
[23:53:53] * BBB has vc1 slices working
[23:53:56] <BBB> that was about time
[23:54:05] <BBB> was getting bored of debugging the same issue over and over again
[23:54:11] <mru> take my brother for instance, he has a phd in maths and still he's only a lowly c++ codemonkey
[23:54:15] <BBB> yet another feature that NOBODY USES finally supported in ffmpeg :-p
[23:54:49] <BBB> mru: don't "hogskola" give out bachelor degrees? the dutch ones do nowadays
[23:55:14] <mru> they give out some degrees, not sure exactly what the rules are
[23:59:28] <BBB> what useless vc1 feature shall I add support for now?
[23:59:33] <BBB> maybe luma/chroma scaling!!

More information about the FFmpeg-devel-irc mailing list