[FFmpeg-devel-irc] IRC log for 2010-03-12
irc at mansr.com
irc at mansr.com
Sat Mar 13 01:00:42 CET 2010
[00:03:17] <mika_video> Hello! I think I just found a bug in function which starts at line 2974 of file libavformat/utils.c : 02974 void dump_format(AVFormatContext *ic, .... actual bug: lines 02980-02982 : uint8_t *printed = av_mallocz(ic->nb_streams); if (ic->nb_streams && !printed) return;
[00:03:19] <mika_video> I think it should return immediately if either (ic->nb_streams == 0) OR (printed == null) // i assume av_mallocz returns null if it fails to allocate memory.
[00:03:24] <mika_video> Should it be rewritten as: if ( (ic->nb_streams == 0) || (printed == NULL) ) return;
[02:10:59] <kierank> it seems the Dolby E API is based on a floppy disk based authentication system
[02:11:29] <mru> floppy disk based auth system???
[02:11:59] <kierank> you put your floppy disk in and it authenticates you to use the api
[02:16:50] <mru> and that's supposed to be secure?
[02:17:04] <mru> well, in the sense that _nobody_ has a floppy drive I guess it might be...
[02:17:11] <mika_video> <kierank>: Very interesting! What if I install windows (2000/XP) into a virtualbox running on linux and mount a diskette image to be seen as drive A: of the vurtual machine? Would the authentication succeed or fail ?
[02:17:51] <kierank> http://www.paceap.com/key.html
[02:18:19] <mika_video> And, can a virtual diskett image of a 3.5" floppy be something else than 80*18*2*512 bytes ?
[02:19:24] <kierank> dunno. all I saw was an error message that said your "key diskette" was invalid to use the dolby e api
[02:19:39] <kierank> and i guess it reflects the current up-to-dateness of dolby
[02:19:40] <mika_video> I know sometimes copy protect systems have used a dloppy formatted in a nonstardard way - so it might not even have the same # of sectors on each track... I have no idea whether it is possible to have a virtual image of such nonstandard floppy ...
[02:19:53] <Dark_Shikari> so apparently I just got an email from google
[02:20:04] <Dark_Shikari> all-but-begging me to apply to work for them this summer
[02:20:50] <kierank> enjoy working on vp8
[02:20:54] <mika_video> <Dark_Shikari>: did you see mu question about a possible bug in libavformat/utils.c ?
[02:20:54] <astrange> apparently?
[02:21:22] <astrange> i didn't make it through their automatic project matcher, maybe they should have bought on2 earlier
[02:21:31] <Dark_Shikari> lol
[02:24:09] <mika_video> does anyone know how olng the avformat_version() function has been available ?
[02:24:49] <CIA-92> ffmpeg: ramiro * r22471 /trunk/doc/developer.texi: Don't use shell syntax to set variables for make.
[02:25:01] <mru> since r14667
[02:25:11] <mika_video> void dump_format(AVFormatContext *ic, .... actual buggy: lines 2980-2982 : uint8_t *printed = av_mallocz(ic->nb_streams); if (ic->nb_streams && !printed) return;
[02:26:41] <mika_video> how does that r14667 splitv into the major.minor.micro -format returned by avformat_version() ?
[02:27:46] <mru> 52.20.0
[02:28:33] <mika_video> is there some way to calculate that ?
[02:28:46] <mru> calculate from what?
[02:29:10] <mika_video> calculate from r14667 to the format 52.20.0
[02:29:53] <mru> look in avformat.h for that revision
[02:31:42] <mika_video> Semms ffmpeg has two totally unrelted version numberings - is that why there's an interesting comment in she source file: "unsigned avformat_version ( void ) // I return the LIBAVFORMAT_VERSION_INT constant. You got a fucking problem with that, douchebag?"
[02:32:10] <Dark_Shikari> lol
[02:32:13] <Dark_Shikari> I love that comment
[02:32:35] <mru> first person pejorative
[02:32:41] <mika_video> I'd like to talkk that into a voice file... just include that as an embedded mp3 resource in ffmpeg.exe
[02:34:26] <Dark_Shikari> hmm, found a random interesting paper on h264 hardware implementations that actual has power usages and other useful stats
[02:34:41] <Dark_Shikari> for a design targeting 720p @ 30fps:
[02:34:53] <Dark_Shikari> 262k gates, 2.44 mm^2, 20ns clock period, 180 milliwatts
[02:35:02] <mika_video> <Dark_Shikari>: what about replacing lines lines 2980-2982 of libavformat/utils.c: uint8_t *printed = av_mallocz(ic->nb_streams); if (ic->nb_streams && !printed) return;
[02:35:07] <mika_video> I think it should return immediately if either (ic->nb_streams == 0) OR (printed == null) // i assume av_mallocz returns null if it fails to allocate memory.
[02:35:16] <Dark_Shikari> one interesting thing is their QCIF version is only like 30k gates less
[02:35:20] <mika_video> Should it be rewritten as: if ( (ic->nb_streams == 0) || (printed == NULL) ) return;
[02:35:21] <Dark_Shikari> and not much slower
[02:35:25] <Dark_Shikari> or simpler
[02:35:40] <Dark_Shikari> which might explain why tons of hardware devices just stick 1080p asics in for the hell of it
[02:39:31] <Dark_Shikari> mru: you at all familiar with how linux handles stack growth>
[02:39:31] <Dark_Shikari> ?
[02:39:58] <mru> unmapped page below stack
[02:40:12] <Dark_Shikari> yes, but what if I have a function that allocates more than one page worth of data?
[02:40:13] <mru> on a page fault it allocates a new physical page and maps it in
[02:40:19] <Dark_Shikari> say, sub esp, 400000
[02:40:23] <Dark_Shikari> it'll jump the page
[02:40:37] <mru> try and see
[02:40:59] <Dark_Shikari> seriously though, I'm just wondering if it does it the same way windows does
[02:41:01] <mru> I think it tries to keep a large unmapped zone between the stack and anything else
[02:41:05] <Dark_Shikari> windows allocates a few guard pages
[02:41:11] <Dark_Shikari> and if a function allocates a _huge_ amount of stack
[02:41:16] <Dark_Shikari> it inserts a call to __chkstk
[02:41:24] <mru> that's the compiler
[02:41:26] <Dark_Shikari> which literally just touches a page at the time
[02:41:27] <mru> gcc doesn't do that
[02:41:29] <Dark_Shikari> mru: it's required by the ABI
[02:41:30] <Dark_Shikari> yes it does
[02:41:38] <mru> hmm, on windows maybe
[02:41:40] <Dark_Shikari> yes, on windows
[02:41:46] <Dark_Shikari> that's how windows does stack handling
[02:41:50] <mru> I know
[02:42:00] <Dark_Shikari> seems like a pretty good method actually
[02:42:09] <mika_video> 52.20.0 - that explains why I get this: avformat_version() not found in avformat-50.dll
[02:42:17] <mru> seems like you're doing something wrong if you need more than 4k in a single stack frame
[02:42:31] <Dark_Shikari> ffmpeg is filled with functions that take 10 times that
[02:42:41] <mru> bugs
[02:42:45] <mru> should be fixed
[02:42:52] <Dark_Shikari> also, large stack sizes aren't exactly unreasoanble
[02:42:53] <mru> where are they?
[02:42:59] <Dark_Shikari> it's more efficient than heap allocation if you intend to re-use the space
[02:43:07] <Dark_Shikari> better on cache.
[02:43:16] <mika_video> why use that much stack space... you could use malloc if you prefer C library functions....
[02:43:43] <Dark_Shikari> (where "large" means 4k or 8k or 32k, not 500k)
[02:44:00] <mru> large is 1k
[02:44:28] <mru> if you get even close to 1k in a single function you should think very carefully about what you're doing
[02:44:52] <mika_video> Has anyone compiled ffmpeg using tcc instead of gcc ?
[02:44:55] <mru> I've worked on systems where 2k for the entire stack was normal
[02:45:10] <mru> mika_video: you joking?
[02:45:21] <Dark_Shikari> mru: 1k isn't enough for the simplest possible dct trellis
[02:45:23] <mru> tcc, toy c compiler
[02:45:24] <mika_video> or at least some demo app using libavformat and libavcodec ?
[02:45:48] <astrange> if you want faster compiles use clang (and a configuration that it doesn't miscompile)
[02:45:53] <Dark_Shikari> lol
[02:45:54] <mru> Dark_Shikari: maybe you shouldn't be using the stack for that
[02:45:57] <Dark_Shikari> mru: why?
[02:46:01] <Dark_Shikari> it's local data which will never be used again
[02:46:11] <Dark_Shikari> why should we waste cachelines on it?
[02:46:13] <mika_video> mru: I just want to know whether tcc will do. I do dot care about code size or speed at all.
[02:46:31] <mru> Dark_Shikari: you'll use it again next time that function is called, no?
[02:46:43] <Dark_Shikari> mru: it'll be long out of cache by then
[02:46:49] <Dark_Shikari> L1, at least
[02:46:52] <mru> so will the huge stack
[02:46:56] <Dark_Shikari> no it won't
[02:46:59] <Dark_Shikari> the stack will be re-used by other functions
[02:47:04] <mika_video> So, if it is bif and fat code and very slow, but it compiles without errors and also does not generate runtime errors or hang the pc, then it is fine.
[02:47:08] <mru> only if they also are badly written
[02:47:09] <Dark_Shikari> with the stack, every single function can share the same data for its tempdata
[02:47:17] <Dark_Shikari> with the heap, you can't share data unless you use scratch buffers
[02:47:23] <Dark_Shikari> in which case you are duplicating the stack on the heap
[02:47:24] <Dark_Shikari> and you are retarded
[02:47:39] <mru> I didn't say you should never do it
[02:47:42] <Dark_Shikari> gcc will always compile better code for local arrays too.
[02:47:59] <mru> but 99% of all cases I've seen where a function used more than 1k of stack space were stupid
[02:48:28] <Dark_Shikari> well lets look at me_refine_bidir
[02:48:39] <Dark_Shikari> 18 Y, U, and V buffers used for cached interpolated data
[02:49:05] <Dark_Shikari> a "visited" array of size 8*8*8 (bitfield)
[02:49:13] <Dark_Shikari> total: ~7.5k
[02:49:14] <mika_video> it's just that for my own reasons, I want to be able to make my own app compile C source. Tcc is available as a dll too, so that's why I like it. And, it's open source and at löeast the code is more understandable to read comared to gcc.
[02:49:28] <astrange> use clang for that too
[02:49:53] <mika_video> <astrange>:clang?
[02:50:07] <astrange> but there's not much reason to move compilers into a library instead of just execing them
[02:51:31] <mika_video> what about exc'ing mingw-gcc directly from a windows GUI app - there's no msys between translating unix-style /src/ffmpeg to C:\mingwproj\src\ffmpeg
[02:52:17] <ohsix> are you trying to get around some license by compiling the code on the fly?
[02:52:28] <Dark_Shikari> JIT license violation ;)
[02:53:15] <mika_video> no. But using modified tcc would be one possibility to make a sanity check when automatically converting .h to other language than C
[02:53:46] <mika_video> for example, to verify structure sizes and offsets of every field of a struct
[02:55:22] <ohsix> but that depends on the compiler and possible compiler specific options in use at the time
[02:56:09] <mru> struct layout is usually defined by the architecture ABI
[02:57:21] <ohsix> i guess; you still can mess with junk though
[02:57:57] <mika_video> I just cannot understand why avformat-50.dll will put the filename at byte offset 180 of AVFormatContext instead of offset 104.
[02:58:47] <mika_video> It also will report 32768 streams in an mpeg file ... probably some odd mismatch between the structure organization in the dll and the header file
[03:00:53] <mika_video> How about revision r15291, 10 Sep 2008. Is that dll still available somewhere ?
[03:04:56] <mru> mika_video: you can build it yourself
[03:05:41] <mika_video> mru: is there anywhere documentation how to do that as a cross-compile, compiling under linux but producing dll for windows
[03:06:41] <mika_video> Since if I have to use gcc to compile, I'd rather use gcc on linux than on windows. But how to get a special version of gcc that runs on linux but compiles windows apps and dll's
[03:07:16] <mru> that's pretty far off-topic here
[03:07:28] <mru> it's possible, that's all I know
[03:08:06] <mika_video> Then just move chat about that into #ffmpeg instead
[03:12:31] <mika_video> And, someone should write a web page explaining all three version numbers: a) the r15291 -style revision number. b) version number like FFmpeg 0.5.1 c) the version number returned by avformat_version()
[03:12:50] <Dark_Shikari> svn version, ffmpeg version, and library version
[03:12:52] <Dark_Shikari> what's so hard about that?
[03:13:13] <mika_video> To understand what all thos numbers actually mean
[03:13:20] <Dark_Shikari> 11:12 <@Dark_Shikari> svn version, ffmpeg version, and library version
[03:15:37] <mika_video> is it the library version = returned by avformat_version(), that an app should check and refuse to run if MAJOR has cahnged but agree to run if MAJOR has stayed the same, even if both MINOR and MICRO have changed ?
[03:16:04] <Dark_Shikari> MINOR is still an api change
[03:16:07] <Dark_Shikari> MAJOR breaks compilation
[03:16:10] <Dark_Shikari> MINOR breaks ABI
[03:16:37] <mika_video> so only MICRO change is ok without recompiling the app ?
[03:16:51] <Dark_Shikari> iirc, yes, that's the only one that's guaranteed
[03:16:58] <mru> MINOR changes are backwards ABI compatible
[03:17:46] <Dark_Shikari> ah ok
[03:17:51] <Dark_Shikari> in that case it's just major that breaks abi
[03:17:55] <mru> yes
[03:18:01] <Dark_Shikari> wait, then why was the padding increase considered a minor bump?
[03:18:03] <mru> minor is increased when new stuff is added
[03:18:06] <mika_video> seems tha mru and Dark_Shikari disagree whether or not a MINOR library version change breaks binary compatibility between the library and an application using it.
[03:18:15] <Dark_Shikari> mika_video: no, it seems that mru corrected Dark_Shikari
[03:18:19] <Dark_Shikari> and stop talking about people in the third person
[03:18:22] <Dark_Shikari> it's really fucking obnoxious
[03:18:24] <mru> Dark_Shikari: the padding thing was a mistake
[03:18:33] <Dark_Shikari> mru: yes, but even so, people commented that there should be a minor bump
[03:18:49] <mru> people were mistaken I guess
[03:19:20] <mru> it's not totally obvious
[03:19:28] <mika_video> "MINOR changes are backwards ABI compatible" - so can I trust this one ?
[03:19:30] <mru> it will still run and do the right thing most of the time
[03:19:40] <mru> yes, unless we make a mistake
[03:30:05] <peloverde> mru: I don't know why but gmail seems to be deciding that all mail coming from you is spam. I can easily fix it for me but others may be missing mail from you.
[03:30:31] <mru> any indication why?
[03:30:33] <ohsix> X-Spamassassin-Cooties: 100
[03:31:21] <mru> sending to myself works fine
[03:38:05] <mru> and either way there's nothing I can do about it
[03:38:19] <mru> if people use a crap mail service, so be it
[03:41:34] <astrange> i keep getting spam with my own email in from: and gmail puts it all in the sent email label
[04:50:04] <justlooking_> kshishkov> what's not pointless then? < perhaps someone capable of making a WMA-Pro Encoder for PS3 Media Server etc http://ps3mediaserver.org/forum/viewtopic.php?f=14&t=5916
[04:51:20] <Dark_Shikari> wmapro encoder? that's retarded
[05:17:32] <CIA-92> ffmpeg: conrad * r22472 /trunk/libavformat/ (6 files): Add ff_ prefix to vorbis_comment
[05:17:32] <CIA-92> ffmpeg: conrad * r22473 /trunk/libavformat/ (6 files): oggdec: Metadata is per-stream; don't merge multiple streams' together
[05:17:32] <CIA-92> ffmpeg: conrad * r22474 /trunk/libavformat/ (oggparsevorbis.c oggdec.h): Const correctness for vorbis_comment
[05:17:33] <CIA-92> ffmpeg: conrad * r22475 /trunk/libavformat/oggparseogm.c: oggdec: Read ogm metadata
[05:17:33] <CIA-92> ffmpeg: conrad * r22476 /trunk/libavformat/oggparseogm.c: oggdec: Cosmetics
[05:17:33] <CIA-92> ffmpeg: conrad * r22477 /trunk/libavformat/oggparsevorbis.c: oggdec: Parse ogm chapters
[05:17:36] <CIA-92> ffmpeg: conrad * r22478 /trunk/libavformat/ (oggdec.c oggdec.h):
[05:17:36] <CIA-92> ffmpeg: oggdec: Pass packets to header() until the stream is done with them
[05:17:36] <CIA-92> ffmpeg: This fixes some old ogm files that had the 3rd vorbis header after a data
[05:17:36] <CIA-92> ffmpeg: packet in another stream. This is invalid in ogg, but this change shouldn't
[05:17:36] <CIA-92> ffmpeg: affect the behaviour of any valid file.
[06:59:19] <pJok> mornings :)
[06:59:29] <av500> gm
[06:59:41] <kshishkov> goda morgnar
[06:59:53] <elenril> o/
[07:01:01] <kshishkov> elentril: [message title="good morning" author="me" performer="me"]
[07:02:20] <elenril> performer shouldn't be set when it's the same as author =p
[07:03:06] <elenril> see also http://tvtropes.org/pmwiki/pmwiki.php/Main/DepartmentOfRedundancyDepartment
[07:24:41] <pJok> author, artist and performer
[07:52:29] <KotH> grüezi wohl
[07:53:06] <kshishkov> shalom
[07:56:23] <av500> slalom
[07:56:35] * kshishkov is not into skiing
[07:59:51] <justlooking_> Dark_Shikari> wmapro encoder? that's retarded < why ? , it serves a good purpose for the many 360 owners around the world, far more tthan say NUT does, sure its would be far better if Microsoft were to just update their firmware with better optimised and capable codecs adn containers but they wont and dont.
[08:00:26] <kshishkov> IIRC, you can have mp3 in asf
[08:00:38] <av500> yes
[08:01:13] <av500> mpeg4 and mp3, it can be the new AVI :)
[08:01:17] <justlooking_> streamable by the 360 ,No, or at least i cant get it playing with ps3ms.
[08:01:35] <justlooking_> steamable and Playable by....
[08:02:14] <kshishkov> speaking what you are compehend I can not
[08:03:24] <justlooking_> ohh i thought you were talking about using mp3 in asf for the 360, sorry....
[08:04:05] <kshishkov> I was
[08:04:05] <av500> since the whole things is about mutlichannel, using mp3 would not solve that isse.
[08:04:12] <kshishkov> ah
[08:04:24] <kshishkov> AAC + XBMC then ;)
[08:05:06] <justlooking_> chiped 360 and a PPC port of linux would be required for that.
[08:06:07] <justlooking_> most people just want to use the broken DLNA client inside the thing.
[08:06:21] <av500> kshishkov: so go write WMAPRO enc!
[08:06:37] <astrange> does wmapro have a max bitrate?
[08:06:50] <kshishkov> av500: oh yeah, go to Ukraine yourself!
[08:06:53] <astrange> it's probably not hard to write a tolerable encoder if you just give up efficiency completely
[08:07:09] <superdump> s/dlna/upnp/
[08:07:13] <superdump> unless something changed
[08:11:11] <kshishkov> and the next thing somebody asks for WMV3 encoder
[08:12:36] <KotH> oh, you're going to write a wmv3 encoder?
[08:12:38] <KotH> cool!
[08:12:39] <KotH> ;)
[08:13:34] <kshishkov> no, you have Dark_Shikari for that
[08:17:04] <justlooking_> that whitefrt guy on the link i posted above wrote wrapper for the "media foundation framework" as use by media center 7 for transcoding AVC/WMA-PRO 5:1.RTP exactly because ffmpeg doesnt do WMA-PRO encoding, its not portable but it works apparently kshishkov , but running XP ps3ms here u cant test it and say for sure.
[08:20:04] <av500> well, transcoding to wmapro using window media api seems straightforward
[08:29:10] <justlooking_> astrange> does wmapro have a max bitrate? < i dont know about a max limit , but MS restrict their VC-1/wmapro 5:1 to 15Mbit/s according to their website data , so they didnt even try and optimise their 360 codecs for their HW.
[08:30:00] <pJok> optimization would mean that they have internal communication between departments
[08:32:16] <superdump> i assume you want wmapro for multi-channel
[08:32:31] <superdump> does the 360 not support any other multi-channel codecs?
[08:33:14] <kshishkov> it's M$ thing
[08:34:46] <justlooking_> yes, thts the ONLY reason anyone would want it OC, id much prefer better 3x3gig PPC optimised codecs/containers in the next firmware update, but MS wont do that, so nect bets thing, find ways to improve what we have.
[08:34:55] <justlooking_> next best thing.
[08:35:59] <superdump> hmm
[08:36:14] <superdump> so it doesn't support multi-channel LPCM?
[08:37:30] <kshishkov> superdump: http://www.engrish.com//wp-content/uploads/2010/03/2rd-floor.jpg
[08:38:21] <justlooking_> AVC/AAC2:1 10Mbit/s or VC1/WMA-PRO5:1 15Mbit/s ,thats about it until whitefrt found it would also do RTP AVC/WMA-PRO5:1
[08:39:54] <justlooking_> in a round about and not portable fasion OC.
[08:42:43] <superdump> kshishkov: i think the l/r swap in pronunciation is quite interesting
[08:42:52] <superdump> but i was speaking with a guy from hong kong a few weeks ago
[08:43:00] <superdump> he said he'd been learning english for 10 years or something
[08:43:13] <superdump> but in the classes, they never speak english to learn pronunciation
[08:43:19] <superdump> they just learn grammar
[08:43:58] <superdump> maybe i'm weird but one of the first things i found myself doing in sweden was learning the sound and rhythm of the language
[08:44:30] <kshishkov> ja, svenska Ã¤r rolig
[08:45:15] <wbs> superdump: how long have you been in sweden?
[08:46:26] <pJok> svenska är rolig
[08:46:29] <kshishkov> longer than me at any rate
[08:46:53] <pJok> om du inte pratar svenska bra som jag, då är det roligt
[08:48:26] <kshishkov> I've heard people speaking in DSB train once, it's like they spoke Swedish worse than me
[08:48:47] <superdump> wbs: i've spent some months there since last year
[08:48:54] <superdump> back in the uk for a couple of weeks
[08:49:00] <superdump> but i'm moving to stockholm when i return
[08:49:16] <wbs> ah, I see
[08:49:43] <kshishkov> wbs: and I've been to Finland for ~1.5 days
[08:50:09] <wbs> kshishkov: grattis ;-)
[08:50:40] <justlooking_> kshishkov> what's not pointless then? < thats the Only reason i posted the WMA-PRO5:1 thing in the first place, but i take it its to hard or fun and you dont want to, so fair enough, i know my typing skills are bad ,but when people are more interested in talking about engrish its time to stop asking.
[08:50:46] <superdump> i was in finland for a couple of days too
[08:50:50] * kshishkov wonders why Finns have town named after the holiest Ukrainian food while Ukrainians don't
[08:51:12] <wbs> kshishkov: which one is that?
[08:51:45] <kshishkov> justlooking_: well, we are too lazy to do that, and some of us have short attention span.
[08:51:49] <kshishkov> wbs: Salo
[08:52:02] <wbs> kshishkov: ah, interesting. :-)
[08:52:11] <wbs> kshishkov: I actually worked in Salo for 1,5 years
[08:52:46] <kshishkov> and I only passed it on a way from Ã
bo to Helsingfors
[08:53:02] <kshishkov> (or from Turku to Helsinki, don't remember)
[08:53:29] <wbs> i prefer åbo and helsingfors :-)
[08:54:03] <kshishkov> well, those names are usually surrounded by the words I understand better ;)
[08:54:24] <pJok> kshishkov, the danes in the dsbfirst trains usually speak very bad swedish
[08:55:02] <pJok> kshishkov, but then again, sometimes they speak swanish on purpose so they don't have to say the same thing twice
[08:55:16] <kshishkov> pJok: it was Fredericia-Copenhagen IC3 but I suspect they spoke Danish
[08:55:45] <kshishkov> pJok: and not even intended to speak Swedish
[08:56:15] <pJok> kshishkov, thats when they speak danish
[08:56:30] <pJok> IC3 trains only have announcements in danish and sometimes english
[08:57:36] <kshishkov> denglish is more likely
[08:58:11] <kshishkov> it's just Danish sounds to me like badly maimed Swedish, written form is even more so
[08:58:17] <pJok> it is
[08:58:36] <pJok> but then again, to me written swedish just simultaneously translates to danish when i read it
[08:58:45] <pJok> i dont do that with german and english
[09:01:47] <kshishkov> looks like Swedish loanwords come from French for better sounding and from English while Danish borrowed from German
[09:02:26] <pJok> actually swedish, danish, norwegian, farose and icelandic are all the same base language
[09:03:04] <kshishkov> yes, Norsk
[09:03:22] <pJok> actually icelandic and farose are closer to the original language
[09:03:55] <pJok> seeing as they are more isolated, so less influence from the outside world
[09:03:55] <kshishkov> yes, being isolated and whatever
[09:04:25] <pJok> also remember that denmark occupied a lot of england
[09:04:33] <pJok> and germany
[09:04:51] <pJok> so the influence is a lot bigger from those languages
[09:04:52] * kshishkov has never heard of Hamlet from Helsingo/r
[09:06:02] <pJok> its still funny that the "swedish mentality" about denmark is that denmark is just another country while the skånska and danish mentality is that skåne and sjælland are just the same region
[09:06:30] <pJok> but then again, copenhagen is next to skåne while stockholm is nearly 600km away
[09:07:10] <kshishkov> yes, Dnes tried to move their capital to the furthest point from the center of Denmark indeed
[09:08:34] <pJok> i dont even notice that i live in sweden and work in denmark
[09:08:51] <av500> mru: ping
[09:09:07] <av500> OBJS-$(ARCH_ARM) += arm/h264pred_init_arm.o
[09:09:25] <kshishkov> pJok: lucky you, here you'll be constantly reminded in what Ukraine you live :(
[09:09:33] <av500> NEON-OBJS-$(CONFIG_H264_DECODER) += arm/h264pred_neon.o
[09:09:54] <av500> if you build for ARM but not H264, it fails
[09:10:35] <kshishkov> disable SVQ3 and RV3/4 as well and it should work
[09:10:53] <av500> they are
[09:11:09] <av500> I have no video decoder enabled, but I build for arm
[09:11:15] <av500> and neon
[09:11:51] <av500> so h264pred_init_arm.c tries to init the neon stuff for h264
[09:12:03] <pJok> kshishkov, maybe it gets better in about 200 years?
[09:12:32] <kshishkov> pJok: unlikely
[09:15:37] <pJok> kshishkov, you are so optimistic :)
[09:16:01] <pJok> but then again, i've never been to the ukraine so i can't really judge on how things are
[09:16:25] <pJok> i've only been to the czech republic and the baltics
[09:17:01] <kshishkov> those are totally different
[09:18:05] <av500> kshishkov: http://pastebin.com/2HsjrxWk
[09:18:09] <pJok> those are the only former semi-russian countries i have been to
[09:18:19] <kshishkov> Kiev (which differs from the rest of Ukraine though) is like Copenhagen with less maintainance of buildings and more drunks
[09:18:54] <pJok> so a bit like prague then
[09:19:02] <pJok> and riga
[09:19:38] <kshishkov> av500: I told you that code is also used by SVQ3 and RV3/4 decoders, CONFIG_H264_DECODER may be not enough
[09:20:03] <av500> right
[09:20:13] <av500> I would rather solve it in the Makefile
[09:20:36] <kshishkov> pJok: Riga was always a piece of Europe under Soviet occupation to me and the only common thing we have with Praha is their old trams ;)
[09:24:10] <pJok> kshishkov, true... and the baltics have both been danish and swedish before they were russian
[09:24:52] <kshishkov> pJok: Russians even illegally built their future capital on Swedish land
[09:25:48] <pJok> well, borders back then were blurry
[09:26:24] <kshishkov> pJok: this may be considered a typical view here - http://commons.wikimedia.org/wiki/File:John_the_Baptist_church.jpg
[09:27:10] <pJok> not an all bad view
[09:27:31] <kshishkov> that's because you don't see all that dirt
[09:27:44] <pJok> its not a street view
[09:27:50] <pJok> i imagine all the dirt being there
[09:37:36] <av500> kshishkov: but then it is still wrong as arm/h264pred_neon.o is only added for CONFIG_H264_DECODER, not for SVQ3 or RV...
[09:38:53] <kshishkov> looks like somebody was no too eager to have faster RealVideo decoding on ARM :)
[09:39:16] <av500> I have it on the DSP :)
[09:40:10] <kshishkov> have what?
[09:40:53] <av500> RV
[09:41:14] <kshishkov> how much buffering DSP interface require?
[09:41:22] <av500> buffering?
[09:42:36] <kshishkov> yes, it's Real it must have buffering
[09:42:48] <av500> :)
[09:42:53] <av500> we did not license that part
[09:44:08] <av500> kshishkov: http://pastebin.com/d8wCdrar
[09:44:24] <av500> that would be needed, no?
[09:45:13] <kshishkov> av500: seems reasonable
[09:45:25] <superdump> does someone have a copy of 14496-3 subpart 1
[09:45:27] <superdump> ?
[09:46:03] <av500> http://www.filestube.com/i/iso+iec+14496+3+pdf
[09:46:46] <av500> ftp://ftp.tnt.uni-hannover.de/pub/MPEG/audio/mpeg4/documents/w2203/w2203.pdf
[09:47:12] <kshishkov> it may be a good thing to have all closed standards bought for FFmpeg devs by our NGO
[09:47:58] <bilboed-tp> kshishkov, that's gonna be ONE BIG BILL
[09:48:49] <superdump> av500: what's the user/pass for that?
[09:48:53] <kshishkov> bilboed-tp, yes and since it's in Swiss Franks we'll leave it to KotH to pay :)
[09:49:08] <bilboed-tp> rofl
[09:49:24] <av500> superdump: for the ftp link?
[09:49:28] <superdump> yes
[09:49:35] <av500> worked for me like that (in a browser)
[09:49:47] <av500> but the ftp seems to be a draft
[09:49:59] <superdump> as long as it's readable, that's ok
[09:50:29] <av500> I could just click on it here: ftp://ftp.tnt.uni-hannover.de/pub/MPEG/audio/mpeg4/documents/w2203/
[09:50:50] <av500> I also have the real one if you want it
[09:52:39] <superdump> i got that pdf
[09:52:43] <superdump> seems rather old
[09:53:03] <av500> wait a sec
[09:55:54] * KotH notes, that he should revive that documents folder on mphq
[10:01:00] <KotH> lu_zero: as i said, deside on what you want and let me know. i'll take care of the rest
[10:02:24] <kshishkov> M$ IIS :)
[11:03:55] <elenril> Dark_Shikari: so our resident theoretical guru said that infalling observer 1)_will_ fall into the singularity in his finite time (and there _will_ be a singularity)
[11:04:15] <elenril> and 2) he won't see any radiation in his frame
[11:04:52] <elenril> for him the black hole doesn't evaporate
[11:05:15] <mru> what happens if schrÃ¶dinger's cat falls into a black hole?
[11:05:41] <iive> it dies for sure.
[11:05:53] <elenril> http://tvtropes.org/pmwiki/pmwiki.php/Main/RocksFallEveryoneDies
[11:09:19] <kshishkov> well, no radiation = nothing to see = useless observer = very short paper
[11:09:36] <elenril> i meant no hawking radiation
[11:10:23] <kshishkov> for large black holes yes
[11:10:32] <elenril> yes what
[11:10:42] <justlooking_> so ask your "resident theoretical guru" explain why is it that any giben mass of clusters or galaxies is always directly proportional to the mass of the central black hole!,
[11:10:45] <kshishkov> no Hawking radiation
[11:11:05] <justlooking_> giben/given
[11:11:47] <kshishkov> justlooking_: maybe because that's 90% of mass :)
[11:12:28] <elenril> kshishkov: no, for any black hole
[11:12:42] <kshishkov> elenril: still, it looks like that falling observer has actually little to observe
[11:13:10] <kshishkov> elenril: so you just don't believe in it?
[11:14:10] <justlooking_> your forgettinf about so called dark flow and dark matter(who comes up with these names), thats said to be 75% of all mass inthe universe in current thinking.
[11:14:25] <elenril> no, only the falling object doesn't see any radiation
[11:14:39] <elenril> the rest of the universe will see it
[11:16:01] <kshishkov> so falling object will be bored to death when it reaches black hole :)
[11:17:56] <kshishkov> mate!
[11:18:18] <kshishkov> any barrel worth looking at?
[11:27:35] <av500> mru: thx
[12:26:46] <CIA-92> ffmpeg: aurel * r22479 /trunk/libavcodec/Makefile: fix compilation with --disable-everything --enable-decoder=twinvq
[12:27:17] * elenril pokes Honoome
[12:27:36] <CIA-92> ffmpeg: aurel * r22480 /trunk/libavcodec/Makefile: fix compilation with --disable-everything --enable-muxer=mpegts
[12:28:21] * KotH pokes elenril
[12:28:54] <Honoome> elenril: yeah?
[12:49:51] <gthreepwood> Hi, I'm having trouble maintaining debug symbols during compile. I'm configuring using the arguments: --enable-debug=1 --disable-optimizations --disable-asm --disable-stripping. I can get function names in a backtrace but not source files or lines.
[12:56:55] <gthreepwood> nvm, --enable-debug=1 was erroneous
[13:04:31] <CIA-92> ffmpeg: ramiro * r22481 /trunk/configure: Document --disable-everything in configure --help.
[13:15:30] <elenril> Honoome: maybe you should ping your patch
[13:20:10] <Honoome> elenril: is baptiste around now? ^^
[13:21:45] <elenril> i think he committed some things yesterday
[13:25:42] <kierank> can ida pro give you access to the memory map like ollydbg can?
[13:46:39] <mru> yay, all the fate systems are back online
[13:47:17] <kshishkov> good
[13:47:24] <mru> and ia64 is failing
[13:47:33] <kshishkov> not so good
[13:47:36] <av500> is that news?
[13:48:23] <mru> it's our old friend: relocation truncated to fit: GPREL22 against `.bss'
[13:51:45] <kshishkov> any fix except random code twiddling?
[13:52:22] <mru> fixing gcc?
[13:52:35] <kshishkov> not an option
[13:56:28] <av500> fixing ia64?
[13:56:38] <kshishkov> it came pre-broken
[13:57:20] <janneg> ignore it, intel seems to do it too
[13:57:33] <Kovensky> at least gcc doesn't run out of registers
[13:58:50] <kshishkov> http://en.wikipedia.org/wiki/Itanic says it has 128 regs, it's hard to run out of all of them even for GCC
[13:59:07] <mru> gcc often performs worse with more registers available
[13:59:36] <kshishkov> first four words are enough
[15:13:05] <CIA-92> ffmpeg: michael * r22482 /trunk/ffplay.c: Print skip_frames variable.
[15:19:00] <gangil> hi ,I want to be a GSOC student with ffmpeg , how can I know which qualification tasks are open ?
[15:21:31] <Dark_Shikari> most of them
[15:22:38] <CIA-92> ffmpeg: michael * r22483 /trunk/libavcodec/h264.c:
[15:22:38] <CIA-92> ffmpeg: Always reset slice_table.
[15:22:38] <CIA-92> ffmpeg: Previously, the area of a lost slice would be left at the slice number of the previous
[15:22:38] <CIA-92> ffmpeg: frame which could occasionally match the number of the next slice and thus a non existing
[15:22:38] <CIA-92> ffmpeg: slice could have been used for prediction leading to additional decoding errors in otherwise
[15:22:39] <CIA-92> ffmpeg: undamaged slices.
[15:23:07] <kshishkov> gangil: claimed tasks are marked as such in our wiki
[15:23:16] <Dark_Shikari> yay, here comes michael's improvements to error concealment
[15:24:16] <CIA-92> ffmpeg: michael * r22484 /trunk/libavcodec/h264.c: Fix indention
[15:24:31] <av500> kshishkov: wiki missed the WMAPRO encoder qualification task :)
[15:25:23] <kshishkov> av500: so do we
[15:35:33] <BBB> hello, my flamy fellow devs :)
[15:35:38] <kshishkov> BBB: do you want to write WMApro encoder?
[15:35:43] <BBB> no
[15:35:54] <BBB> I'd rather encode to vorbis than to wmapro
[15:35:54] <Dark_Shikari> why in the world would anyone want to do that
[15:35:59] <Dark_Shikari> wmapro is just a ripoff of aac
[15:36:03] <BBB> that too
[15:36:04] <BBB> anyway
[15:36:05] <Dark_Shikari> might as well write a good aac encoder
[15:36:07] <Dark_Shikari> and aac is more useful
[15:36:10] <BBB> I'd like to focus on RE'ing new decoders
[15:36:15] <BBB> decoding is useful
[15:36:18] <Dark_Shikari> WVP2
[15:36:23] <BBB> encoding is like submitting your soul to the devil
[15:36:27] <BBB> yes, WVP2 is high on my list
[15:36:31] <BBB> I'd like to learn video, still
[15:36:37] <kshishkov> Dark_Shikari: some losers with ybox and no desire to install something like XBMC
[15:36:46] <Dark_Shikari> kshishkov: xbox supports aac
[15:36:47] <Dark_Shikari> and ac3
[15:37:04] <kshishkov> not multichannel
[15:37:15] <Dark_Shikari> it supports multichannel ac3
[15:37:42] <kshishkov> whatever, someone asked for it this morning
[15:37:51] <Dark_Shikari> I saw, and it's still stupid
[15:38:10] <Dark_Shikari> writing encoders for proprietary formats should be very carefully considered
[15:38:24] <kshishkov> and rejected freely :)
[15:38:29] <Dark_Shikari> yes, exactly
[15:38:34] <Dark_Shikari> because writing an encoder can be a _Bad_ thing
[15:38:45] <peloverde> we have an alac encoder
[15:38:56] <Dark_Shikari> yeah, but that made sense because:
[15:39:01] <mru> alac isn't a terrible format
[15:39:02] <Dark_Shikari> a) nobody will ever use alac in place of flac when they don't need to
[15:39:09] <mru> it's quite good actually
[15:39:11] <Dark_Shikari> b) ipods will never support flac
[15:39:20] <Dark_Shikari> c) we did a better job than apple so lol
[15:39:26] <Dark_Shikari> d) it was easy (shares a lot of flac code)
[15:39:31] <peloverde> e) stuff doesn't get "stuck" in a lossess format
[15:39:35] <Dark_Shikari> And yeah, that
[15:40:07] <av500> f) lossless does not need an elaborate psych model, no?
[15:40:37] <kshishkov> av500: along with rate control and other mindboggling stuff
[15:40:47] <Dark_Shikari> >implying ratecontrol is hard
[15:41:15] <kshishkov> ratecontrol with proper bit allocation
[15:41:29] <Dark_Shikari> ratecontrol doesn't allocate bits, it chooses quantizers
[15:41:49] <Dark_Shikari> bit allocation is a consequence of ratecontrol, not a a responsibility
[15:42:26] <kshishkov> then why FFmpeg does not have generic mpegvideo-independent rate control yet?
[15:42:47] <peloverde> but sometimes there are external requirements on bit allocation
[15:42:59] <Dark_Shikari> kshishkov: for the same reason it doesn't have a generic psymodel
[15:43:07] <Dark_Shikari> also, mpegvideo has the VBV model to deal with
[15:43:10] <Dark_Shikari> which is the hard part
[15:43:17] <Dark_Shikari> not sure how that comes into play in audio
[15:43:26] <Dark_Shikari> also, audio generally has less rate variation than video
[15:43:30] <Dark_Shikari> and the quantizer scale is different
[15:43:50] <BBB> guys
[15:43:54] <BBB> you're forgetting the whole point
[15:43:58] <BBB> I know nothing about audio
[15:44:00] <Dark_Shikari> no, I have a sharp pencil right here
[15:44:00] <BBB> let alone video
[15:44:05] <BBB> I know the bitstream format
[15:44:07] <BBB> that's about it
[15:44:08] <Dark_Shikari> Then learn
[15:44:20] <BBB> I should not focus on encoding a crappy format with a crappy encoder
[15:44:24] <BBB> it's wrong, double wrong
[15:44:35] <BBB> I should focus on decoding, which is easy, or encoding with good encoders into good formats
[15:44:35] <kshishkov> BBB: I did, so what?
[15:44:43] <BBB> kshishkov: dunno ;)
[15:44:50] <BBB> anyway, no, I won't write a wmapro encoder
[15:44:57] <BBB> but if someone pays me $100k I'll consider it
[15:45:10] <kshishkov> BBB: you'll get many thanks from Diego if you write VoxWare decoder
[15:45:25] <BBB> I thought you wanted me to do intel audio 4?
[15:45:38] <Dark_Shikari> Do wvp2
[15:46:02] <kshishkov> there was only intel audio 2
[15:46:35] <kshishkov> and yes, someone should do wvp2 eventually
[15:46:48] <BBB> Dark_Shikari: how much for wvp2?
[15:46:59] <BBB> ;)
[15:48:23] <kshishkov> BBB: probably a free Facebook account ;)
[15:52:26] <BBB> lol :)
[15:52:58] <bilboed-tp> only if in mint condition
[15:53:18] <av500> I thought a free gmail acount soon, no?
[15:54:47] <BBB> facebook was sponsored by MS
[15:54:52] <BBB> so more likely a free MS Live account
[15:54:56] <BBB> or did they rename it again?
[15:55:09] <BBB> HotMail 4.0
[15:55:44] <kshishkov> a reward from M$ for REing M$ codec? 512 years in jail more likely
[15:57:27] <Dark_Shikari> MS would _loooove_ an open source VC-1 encoder
[15:57:51] <Dark_Shikari> 'course they would have liked it more 5 years ago when it was actually relevant
[16:00:12] <peloverde> then maybe they should open source their encoder
[16:02:10] <Dark_Shikari> >microsoft
[16:02:12] <Dark_Shikari> >open sourcing things
[16:02:13] <Dark_Shikari> ahhaha
[16:02:43] <av500> pwszzstrLicenceDetails
[16:02:49] <kshishkov> but they do it
[16:05:25] <peloverde> they shouldn't be dissatisfied with the lack of an oss encoder if they aren't willing to help
[16:07:58] <kshishkov> we almost got vp6 encoder though :(
[16:13:22] <BBB> kshishkov: do you remember how in ida I get rid of this balloon iew and go back to regular disassembly view?
[16:13:40] <BBB> I keep getting stuck in the arrow-view instead of disassembly view, and don't know how to get out of it
[16:13:40] <kshishkov> right-click menu -> "text view"
[16:13:50] <BBB> ah, that was it
[16:13:58] <BBB> you are my hero
[16:14:03] <kshishkov> :)
[16:14:19] * DonDiego hopes he is nobody's hero...
[16:14:46] <kshishkov> not married, unemployed, has not got degree, third-world country citizen is your hero?
[16:15:11] <BBB> absolutely
[16:15:39] <DonDiego> who *does* have a degree around here? :)
[16:15:59] <BBB> kshishkov has a PhD right?
[16:16:08] <kshishkov> nope
[16:17:00] <kshishkov> it would require me anouther year of useless postgrad work, writing useless thesis and three years of slavery afterwards
[16:18:39] <BBB> hm...
[16:18:46] <BBB> I'll get a PhD in a couple of months, hopefully
[16:21:22] <av500> that means, soon you are the most qualified! :)
[16:22:47] <Dark_Shikari> um, I almost have a bachelor's...
[16:22:48] <Dark_Shikari> ;)
[16:23:02] <BBB> overqualified, really
[16:23:07] <BBB> it doesn't take a PhD to RE codecs
[16:23:12] <BBB> in fact, anyone can do it
[16:23:24] <BBB> a PhD is just a sad excuse for companies to not hire me because they'll think I'm more expensive
[16:23:52] <av500> Phd in biology -> zoo keeper?
[16:24:19] <DonDiego> BBB: try google, they hire a lot of phds
[16:24:39] <BBB> my sis-in-law works at google
[16:24:43] <BBB> not a happy place, might I say
[16:25:00] <BBB> it might've been great 10 years ago, but right now I don't see good vibes
[16:27:56] <kshishkov> BBB: I got my master degree during REing RV4
[16:31:28] <DonDiego> go figure, i thought google was a nice workplace from all i heard..
[16:31:52] <kshishkov> it's mostly rumours what you hear anyway
[16:32:57] <kshishkov> a long time ago Bell Labs was an incrediable place to work at
[16:36:02] <BBB> kshishkov: well, she works in mobile, and basically they work 6hrs/day for seriously long periods of time, and half of the projects get cut then
[16:36:08] <BBB> it's very, very competitive
[16:36:19] <BBB> 16hrs/day, not 6
[16:38:32] <superdump> i was going to say, 6hrs a day seems nice :)
[16:38:37] <KotH> does anyone know where one could get the exact mechanical dimensions of a 3.5" harddrive (aka a mechanical drawing) ?
[16:38:50] <kshishkov> google.com
[16:39:29] <KotH> i already googled
[16:39:33] <KotH> didnt find anything
[16:40:38] <av500> your hdd drawer and a ruler?
[16:41:19] * kshishkov has FDD drive lying around
[16:44:41] <KotH> av500: we need it for production, so we need to have more or less reliable data
[16:52:28] <DonDiego> Yuvi: i assigned you a bunch of bugs, i hope you don't mind..
[16:55:05] <DonDiego> peloverde: did i already pour kind words of thanks, appreciation and professional esteem over you?
[16:55:13] <DonDiego> (for sbr support in aac)
[16:55:42] <peloverde> It's possible, I don't remember :)
[16:55:53] <DonDiego> ok, remind me to do it some day
[16:56:00] <DonDiego> ;-p
[16:56:14] * DonDiego hails peloverde
[16:56:23] * DonDiego hails peloverde some more
[16:56:23] <peloverde> thanks
[16:56:48] * DonDiego hails peloverde until his arms get tired and his throat is sore
[16:57:08] <DonDiego> i'm superhappy :)
[16:57:10] <kshishkov> you'll die praising him when he makes ffaacenc usable
[16:57:36] <Dark_Shikari> We should set up a donation fund for ffaacenc
[16:57:38] <Dark_Shikari> seriosuly
[16:57:41] <Dark_Shikari> I know people who would donate
[16:57:44] <peloverde> thanks again, so know that sbr landed what's the status on 0.6
[16:57:52] <Dark_Shikari> my idea: make it a bounty fund
[16:58:07] <Dark_Shikari> anyone who can make ffaacenc beat LAME in a hydrogenaudio test at 96kbps
[16:58:11] <Dark_Shikari> gets all the money
[16:58:38] <DonDiego> peloverde: that was a big bump in the road for 0.6
[16:58:50] <elenril> now to merge -mt ;)
[16:58:52] <DonDiego> now it depends on Yuvi and his theora and ogg work
[16:59:01] <DonDiego> elenril: if only there was a patch :(
[16:59:09] <DonDiego> astrange: how about proposing it?
[16:59:15] <Dark_Shikari> seriously, merge mt
[16:59:19] <Dark_Shikari> this will make everyone more awesome
[16:59:22] <DonDiego> yes, seriously
[16:59:29] <Dark_Shikari> the problem right now is that it's a maintenance disaster
[16:59:33] <Dark_Shikari> astrange has to keep updating his fork
[16:59:50] <DonDiego> well, propose it and have everybody review it
[17:00:02] <DonDiego> this stuff will generate interest just like aac sbr
[17:00:11] <DonDiego> not only michael reviewed that
[17:00:13] <CIA-92> ffmpeg: michael * r22485 /trunk/libavcodec/error_resilience.c: Fix guess_mv() so that it works correctly with 4x4 MV blocks.
[17:00:34] <DonDiego> Yuvi: can you give me a status update about your theora and ogg work?
[17:00:50] <av500> work on making it go away?
[17:01:24] <DonDiego> lol, no working on making it rock :)
[17:03:44] <DonDiego> Yuvi: i understood r22463 to fix seeking in ogg theora files, but it did not in my tests with mplayer..
[17:03:54] <Dark_Shikari> mplayer doesn't use lavf
[17:04:04] <Dark_Shikari> iirc
[17:04:12] <DonDiego> it does if you tell it to :)
[17:04:12] <mru> it can do
[17:04:34] <DonDiego> and depending on the container lavf is the default, mov/mp4 for example
[17:07:06] <Kovensky> <@DonDiego> lol, no working on making it rock :) <-- you expect him to make ogg rock? o_O
[17:07:53] <mru> some people believe in miracles
[17:08:18] <kshishkov> DonDiego: someone should bury default MPlayer demuxer - it spends a minute trying to guess MPEG format out of some file supported by lavf (usually it's TTA or Bink)
[17:08:29] <jai> the fsf already has a PlayOgg campaign, start a FlayOgg to keep things even
[17:08:30] <DonDiego> i have high hopes of him making the ogg *support* within ffmpeg rock
[17:08:32] <mika_video> is there either a rule or a list, how to convert release numbers like r22485 into either library version numbers or nightly snapshot dates ?
[17:08:48] <DonDiego> kshishkov: give me some examples
[17:09:03] <DonDiego> kshishkov: it should just be a matter of switching defaults for some files
[17:09:07] <KotH> mika_video: yes, using svn and some bash magic
[17:09:11] <Kovensky> mika_video: checkout revision, extract version number from headers? o_O
[17:09:50] <kshishkov> DonDiego: playing any large TTA or Bink file over NFS
[17:09:56] <kshishkov> on a PowerPC
[17:09:59] <DonDiego> not locally?
[17:10:44] <kshishkov> with TTA it's even locally
[17:11:43] <DonDiego> try overriding the demuxer by file extension
[17:11:49] <DonDiego> for tta files
[17:12:04] <DonDiego> if that works, commit
[17:12:11] <kshishkov> why not make MPlayer detect it with lavf first?
[17:12:48] <DonDiego> that's what i'm saying
[17:13:03] <kshishkov> it=everything, not just some exceptions
[17:13:35] <DonDiego> discuss that on the ml
[17:15:12] <Kovensky> <@kshishkov> why not make MPlayer detect it with lavf first? <-- mplayer first tries with lavf, but only sticks with it for the formats on the preferred list
[17:15:16] <Kovensky> then it tries everything else, and if all else fails, lavf again ._.
[17:17:54] <DonDiego> peloverde: what's the status of aac and sbr? any known bugs or regressions?
[17:22:41] <BBB> submit -mt
[17:22:49] <BBB> I'll do initial reviews so michael doesn't have to
[17:23:05] <BBB> review will say: "awesome! please commit!"
[17:24:06] <Kovensky> lol
[17:28:52] <CIA-92> ffmpeg: diego * r22486 /trunk/ (libavdevice/bktr.c configure): (log message trimmed)
[17:28:52] <CIA-92> ffmpeg: Add _XOPEN_SOURCE=600 to CPPFLAGS on NetBSD.
[17:28:52] <CIA-92> ffmpeg: This make NetBSD compile out of the box at the cost of a hack, but a
[17:28:52] <CIA-92> ffmpeg: locally contained one that is preferable to the one used in the wild.
[17:28:52] <CIA-92> ffmpeg: Currently the NetBSD FATE box adds -D_NETBSD_SOURCE to CPPFLAGS to work
[17:28:53] <CIA-92> ffmpeg: around broken system headers. Since it is unlikely for NetBSD to fix their
[17:28:54] <CIA-92> ffmpeg: headers, it is better to use a standard flag instead of a system-specific one.
[17:30:09] <CIA-92> ffmpeg: diego * r22487 /trunk/libavdevice/bktr.c:
[17:30:09] <CIA-92> ffmpeg: Replace BSD-specific integer types by their POSIX equivalents.
[17:30:09] <CIA-92> ffmpeg: Tested to work by Michael Kostylev on the NetBSD FATE box.
[17:30:20] <BBB> DonDiego: ... and add av_cold to init ...
[17:30:27] <BBB> (if you can modify the commit msg)
[17:30:49] <DonDiego> oh, crud
[17:30:53] <DonDiego> that slipped in..
[17:31:05] <DonDiego> i love being able to edit commit messages..
[17:33:48] <peloverde> DonDiego: no known bugs or regressions
[17:34:12] <DonDiego> ok, cool
[17:34:18] <BBB> TY
[17:34:23] <DonDiego> ty?
[17:34:29] <peloverde> There were a few but they've all been fixed
[17:34:40] <DonDiego> excellent
[17:34:43] * DonDiego pets peloverde
[17:35:02] <DonDiego> don't you dare buy drinks at the next fosdem beer event
[17:35:15] <BBB> ty=thankyou
[17:35:52] <DonDiego> i'll make sure you get your orders
[17:36:08] <DonDiego> served to you by my humble self
[17:36:40] <DonDiego> or others that wish to join me in providing your liquid supply..
[17:36:59] <peloverde> I'm sure by next fosdem there will be a whole new crowd of FFmpeg heroes
[17:37:27] <DonDiego> we'll see :)
[17:37:55] <DonDiego> i'm not sure i will have occasion to buy you beer before next fosdem..
[17:38:01] <av500> there will be kshishkov WMAPRO encoder :)
[17:38:29] <DonDiego> av500: btw, did i see a patch from you being committed?
[17:41:01] <av500> yes
[17:41:10] <av500> a wee tiny one
[17:41:28] <DonDiego> leave the channel
[17:41:32] <DonDiego> immediately!
[17:41:34] <DonDiego> ;)
[17:41:40] <DonDiego> but rejoin right away
[17:42:27] <DonDiego> there you are :)
[17:42:29] <elenril> av500++
[17:42:46] * av500 blushes
[17:43:13] <DonDiego> now keep those patches coming :)
[17:44:21] <av500> I already have a love-hate relationship with the asf muxer... :)
[17:44:44] <elenril> chapters muxing soon?
[17:45:13] <av500> elenril: I'd love to have the negative ts issue resolved 1st
[17:46:06] <av500> and I also want to fix the complaints that M$ ASF viewer has with the files we generate
[17:46:15] <BBB> Vitor1001: btw, since all other filters are finished now, have you had any thoughts about the frequency domain filter in wmavoice yet? If not, I'll start working on that (will require tons of reading)
[17:48:55] <elenril> av500: doesn't the patch i sent today fix it?
[17:49:17] <av500> err which one
[17:49:25] <elenril> riff: don't pad extradata when writing ASF
[17:51:24] <av500> no, not related
[17:51:49] <elenril> then what is the issue?
[17:52:29] <av500> asf viewer complains that some packets are marked as multipayload, but have only 1 payload
[17:52:39] <elenril> oh
[17:52:42] <av500> I think this is a minor thing, but you never know with M$ SW (WMP)
[17:53:00] <elenril> true
[17:53:09] <av500> also, atm it muxes negative timestamps which cannot be stored in ASF
[17:53:25] <av500> hence my quest to not have negative ts...
[18:03:52] <Dark_Shikari> too many fucking companies holy shit
[18:04:03] <Dark_Shikari> I just got the 9th company wanting to potentially hire me for the summer
[18:04:12] <Dark_Shikari> the rate is increasing
[18:04:16] <elenril> they got hit with a clue stick?
[18:04:34] <peloverde> Even I found work, the economy really is turning around
[18:04:47] <Dark_Shikari> autodesk
[18:05:04] <Dark_Shikari> so now the total list of companies that have inquired/offered/whatever relating to jobs:
[18:05:13] <kshishkov> Dark_Shikari: make your new contribution appear under name of, say, Garrett Jason-Glaser ;)
[18:05:24] <Dark_Shikari> gaikai, otoy, zencoder, google, avail, green hills software, autodesk, skype, quallcomm
[18:05:31] <Dark_Shikari> the fuck is with this.
[18:06:37] <elenril> 1) found Dark Shikari Inc
[18:06:42] <elenril> 2) accept all the jobs
[18:06:46] <elenril> 3) ???
[18:07:08] <kshishkov> 3) outsource them to Jai
[18:07:11] <kshishkov> 5) profit!
[18:07:13] <Dark_Shikari> lol
[18:07:17] <Dark_Shikari> these are all onsite jobs though :/
[18:07:20] <Dark_Shikari> harder to outsource
[18:07:52] <kshishkov> mail order Jai and disguise him a bit, problem solved
[18:08:23] <kierank> is .ua not considered far enough to be called "outsourced" yet ;)
[18:08:34] <bilboed-tp> find minions, accept all contracts that offer IRC connectivity, send out minions as yourself
[18:08:35] <bilboed-tp> done
[18:09:44] <kshishkov> kierank: it is, once I had to work disguised as another man in a project outsourced in the fourth grade
[18:10:04] <kierank> "Dave from AT&T" ?
[18:10:17] <kshishkov> in the fourth grade = one company outsourcing it to another company which outsourced it again, continue
[18:10:21] <kshishkov> no
[18:10:23] <kierank> llol
[18:10:42] <kshishkov> it was for some British company at the end
[18:11:09] <kierank> lots of british companies are going to eastern europe now
[18:12:43] <kshishkov> that was 3-4 years ago
[18:13:10] <DonDiego> ffplay keeps crashing for me on os x, can anybody confirm this behavior?
[18:13:32] <DonDiego> also, does anybody know how to turn off the crash report notifications to apple (for ffplay)
[18:13:32] <kshishkov> probably not
[18:13:38] <peloverde> ffplay crashes or hangs for me just about everywhere
[18:14:15] <DonDiego> i cannot really make any sensible tests here.. :-(
[18:14:27] <av500> gdb ffplay?
[18:14:36] <gangil> hi , is there any starting material that can get me started over with the code base ? (just wanted to confirm)
[18:15:10] <av500> grab a random bug and try to fit it
[18:15:34] <DonDiego> read the developer docs and try the multimedia wiki
[18:16:14] <peloverde> gangil: is there something particular you are interested in?
[18:16:18] <DonDiego> btw, we need some benchmarking instructions in the dev docs
[18:16:32] <DonDiego> i'm never sure how exactly to do them
[18:16:58] <kshishkov> ffmpeg -benchmark -i infile -f null
[18:16:59] <peloverde> we also need to update MOOBEX
[18:17:47] <gangil> peloverde: basically , I am trying to get started with the code base , as I want to participate with ffmpeg in GSOC
[18:20:18] <peloverde> codecs are in libavcodec/, (de)muxers are in libavformat/, common code is in libavutil or libavcodec, files are generally named by format name, for things that are shared you can find those by looking at something that uses them
[18:20:47] <DonDiego> gangil: check out the multimedia wiki, there are a few short howtos there
[18:20:57] <peloverde> documentation is at http://ffmpeg.org/documentation.html and in wiki.multimedia.cx
[18:21:06] <DonDiego> http://wiki.multimedia.cx/ there is a category named "ffmpeg", look at that
[18:39:00] <sancode> hello
[18:42:58] <BBB> gangil: whch project will you participate in?
[18:43:43] <gangil> havent decided yet , but I like the organization , as I want to work on codecs etc.
[18:44:04] <gangil> I was a GSOC student last year
[18:44:08] <gangil> with tux4kids
[18:44:28] <BBB> ok
[18:44:34] * gangil successfully completed that
[18:44:38] <BBB> so you might want to look at our proposed project list
[18:44:48] <BBB> because particular projects will have suggested qualification tasks
[18:45:00] <BBB> see http://wiki.multimedia.cx/index.php?title=FFmpeg_Summer_Of_Code_2010
[18:47:19] <gangil> BBB: thanks , I googled it though :)
[18:48:18] <BBB> good :)
[18:48:30] <BBB> a codec I'd be willing to mentor is WVP2, but there's some others as well
[18:48:54] <BBB> you can also suggest your own codec, but be aware that codecs are difficult, so you'll want a good mentor to help you in th einitial stages
[18:49:24] <BBB> I'm not saying you're not competent, I just know how difficult it is ;)
[18:52:07] <gangil> hmm...
[19:08:06] <Vitor1001> BBB: sorry, but I don't think I'll be able to work on it very soon :(
[19:23:05] <BBB> ok, I'll go at it then
[19:23:37] <gangil> BBB: where can I get more info abt this project Implement a Windows Screen Video 2 decoder and can you elaborate onthe qualification task : reverse engineering the frame header ?
[19:24:08] <BBB> merbanan would be a great mentor for that
[19:24:13] <BBB> we used to have a wiki page
[19:27:37] <BBB> hm, I cannot find the wiki page :(
[19:27:43] <BBB> kshishkov: would you be willing to mentor?
[19:27:46] <gangil> the video listed their looks very much the same as created by windows movie maker
[19:27:47] <BBB> kshishkov: that way at least you get $500
[19:28:37] <kshishkov> BBB: unlikely, I'm going to spend that time trying to get out of here. I'll try to help though
[19:32:02] <BBB> ok
[19:32:21] <BBB> gangil: ok, I guess we'll have to wait for either merbanan, or I'll have to do it... vitor is too busy also, it seems :(
[19:32:30] <BBB> gangil: can you email Benjamin Larson?
[19:32:39] <BBB> google his name + ffmpeg to get his email addy
[19:32:45] <BBB> ask him if he'd be willing to mentor
[19:32:52] <BBB> if not I'll help you get set up
[19:33:25] <gangil> ok
[19:34:28] <BBB> sorry we don't have a mentor straight available
[19:34:50] <kshishkov> it does not matter much though
[19:35:00] <gangil> thanks :)
[19:35:03] <BBB> in the mean time, I'd suggest that you get ida pro installed... google it, a company sells it, they have an old version (3.7 or so) available as a free download
[19:35:37] <BBB> then using ida pro, open the MS .dll file that is used to decode wvp2 files, you can find that e.g. in the mplayer dll collection
[19:35:50] <BBB> and mplayer will also serve as a good positive control to actually compare the output against
[19:35:53] <kshishkov> it's wmvadvd.dll
[19:35:56] <drv> http://www.hex-rays.com/idapro/idadownfreeware.htm
[19:35:57] <BBB> that should get you a starting point
[19:36:44] <BBB> ocne you've got it all set up and in front of you, you can start asking questions on where to start, it'll be a bit painful in the beginning but you'll get it quite quickly
[19:36:49] <BBB> hopefully ;)
[19:38:27] <gangil> Benjamin Larson tamom7rt at yahoo.ca
[19:38:42] <drv> it's larsson
[19:38:44] <gangil> this one seems to be it
[19:38:58] <kshishkov> nope
[19:38:58] <BBB> Larsson maybe
[19:39:23] <kshishkov> and domain is .se
[19:39:24] <BBB> banan @ ludd.ltu.se
[19:40:17] <gangil> Benjamin LarssonÂ <banan at ludd.ltu.se >
[19:40:22] <gangil> kk
[19:40:25] <BBB> :)
[19:40:49] <BBB> so what was your last year's project?
[19:41:52] <gangil> it was regarding networking , mainly involving sockets SDL_net http://socghop.appspot.com/gsoc/student_project/show/google/gsoc2009/tux4kids/t124024972814
[19:42:14] <gangil> and C
[19:42:19] <gangil> ofcourse
[19:55:49] <gangil> everyone suddenly got silent .. ?? or am I offline :S
[19:56:10] <kshishkov> maybe you are
[19:56:29] <kshishkov> or maybethere's nothing to discuss
[19:58:48] <sancode> Hello. Who can check our software for LGPL-compliance and give any recommendations..
[19:59:15] <kshishkov> you can do it yourself, for example
[20:02:12] <Dark_Shikari> sancode: you can ask
[20:02:23] <Dark_Shikari> explain your situation and what you've done
[20:02:53] <sancode> our program is FVD Suite. It uses LGPL FFmpeg. So no non-free libraries/codecs
[20:03:04] <sancode> site is : http://flashvideodownloader.org/
[20:03:13] <sancode> it uses ffmpeg like extension for encoding video
[20:03:17] <Dark_Shikari> say more useful things, like
[20:03:23] <Dark_Shikari> "our product is proprietary"
[20:03:27] <Dark_Shikari> or "our product is GPL" or
[20:03:27] <Dark_Shikari> ...
[20:03:33] <Dark_Shikari> and how you are distributing ffmpeg and so forth
[20:03:46] <sancode> closed-sources. FFmpeg bundled with installer
[20:03:54] <sancode> EULA has information about it
[20:03:56] <Dark_Shikari> ffmpeg.exe, or ffmpeg libraries?
[20:04:01] <sancode> ffmpeg.exe
[20:04:10] <sancode> source code and build instructions also available
[20:04:13] <Dark_Shikari> OK, first of all, in that case, there is no difference between LGPL and GPL ffmpeg (basically)
[20:04:15] <sancode> http://flashvideodownloader.org/mirrors/
[20:04:19] <Dark_Shikari> so there is no reason why you can't include a GPL version
[20:04:35] <Dark_Shikari> (and, given that x264 is GPL, and H.264 encoding is rather important for things like ipods, that might be a useful thing to do)
[20:04:44] <BBB> lol :)
[20:04:55] <Dark_Shikari> thus:
[20:04:57] <BBB> sancode: did you check our license recommendations?
[20:05:01] <Dark_Shikari> 1) Does your product include source code?
[20:05:03] <sancode> build parameters
[20:05:04] <sancode> FFmpeg version SVN-r21487, Copyright (c) 2000-2010 Fabrice Bellard, et al.
[20:05:04] <sancode> built on Feb 4 2010 00:11:22 with gcc 3.4.5 (mingw-vista special r3)
[20:05:04] <sancode> configuration: --extra-cflags=-fno-common --disable-shared --enable-static --d
[20:05:04] <sancode> isable-doc --disable-ffplay --disable-ffserver --enable-memalign-hack --enable-b
[20:05:04] <sancode> zlib --enable-libmp3lame --disable-network --enable-swscale --enable-avisynth
[20:05:05] <Dark_Shikari> in the downloader
[20:05:14] <BBB> http://ffmpeg.org/legal.html
[20:05:15] <sancode> product not includes but includes link to source
[20:05:20] <sancode> and all credits/COPYING
[20:05:22] <Dark_Shikari> sancode: link to your website?
[20:05:22] <sancode> with folder of ffmpeg
[20:05:25] <Dark_Shikari> ok, that's good
[20:05:28] <sancode> http://flashvideodownloader.org/
[20:05:34] <Dark_Shikari> Does the product include the (L)GPL?
[20:05:37] <sancode> http://flashvideodownloader.org/fvd-suite/notes.php
[20:05:45] <sancode> iyes
[20:05:55] <Dark_Shikari> anyways, it looks like you're, from a superficial perspective, good
[20:05:57] <Dark_Shikari> this is not legal advice
[20:05:59] <Dark_Shikari> IANAL
[20:07:20] <BBB> the links to ffmpeg.org, license info etc. on the download page are missing
[20:07:33] <sancode> yes i know
[20:07:50] <sancode> web-designer will be only after weekend
[20:08:00] <BBB> well, that's one :)
[20:08:00] <sancode> we will fix it
[20:08:03] <BBB> ok
[20:09:17] <sancode> besides ffmpeg
[20:09:19] <sancode> we distribute
[20:09:32] <sancode> CREDITS MAINTAINERS COPYING
[20:09:38] <sancode> Build instructions (our Readme)
[20:09:44] <sancode> where to download sources
[20:09:46] <sancode> and instructions
[20:10:14] <sancode> about dialog has information of ffmpeg too..
[20:10:20] <sancode> We plan to redisign it
[20:10:44] <sancode> Now just: This products contains FFmpeg <link>
[20:10:53] <sancode> next will be more informative
[20:11:03] <gangil> /j #xelerance
[20:11:17] <sancode> /j #xelerance
[20:11:19] <sancode> sorry
[20:11:41] <gangil> sorry
[20:12:31] <sancode> EULA is http://dl.dropbox.com/u/1811883/license.txt
[20:20:40] <BBB> the about dialog, eula and website should all say "this product contains ffmpeg <link> under the license <license + link>"
[20:20:51] <BBB> and then some way for users to figure out where to get the sources you used
[20:20:58] <sancode> ok
[20:21:04] <BBB> doesn't have to be very wordy
[20:21:07] <BBB> just as long as it's clear
[20:21:34] <BBB> e.g. "this product contains <a href=..>ffmpeg</a> under the <a href=..>LGPLv2.1</a> (<a href=..>sources</a>)
[20:21:34] <BBB> "
[20:21:45] <BBB> that would be enough for the about dialog already
[20:21:59] <BBB> as long as A) credits, B) license and C) sources used are all obvious
[20:22:08] <sancode> good
[20:22:50] <ohsix> is there an advertising clause? these are things that are nice to do but nobody is bound by that stuff
[20:23:06] <sancode> i didn't understand
[20:23:39] <BBB> ohsix: there is a credit clause, yes
[20:23:52] <BBB> advertising is generally interpreted as "you cannot remove this message"
[20:23:58] <BBB> which is lgpl/gpl-incompatible
[20:24:03] <ohsix> where
[20:24:35] <sancode> we are not using FFmpeg name for advertising
[20:24:56] <BBB> 7b
[20:25:09] <BBB> sancode: he doesn't mean advertising in that way ;)
[20:25:23] <BBB> ohsix: let me quote
[20:25:31] <BBB> ohsix: "Give prominent notice with the combined library of the fact that part of it is a work based on the Library, and explaining where to find the accompanying uncombined form of the same work."
[20:25:59] <BBB> ohsix: we interpret prominent as "it should be clearly visible, regardless of method of installation, use and acquisition, to people installing, using and downloading"
[20:26:28] <BBB> ohsix: which means it should be in the program during running ("about dialog"), it should be visible during installatio ("in eula"), and it should be visible during download ("on website download page")
[20:26:44] <BBB> ohsix: your lawyer might suggest something else, but if we're unhappy with your chosen solution, we'll sue your pants off :)
[20:26:51] <ohsix> heh
[20:27:20] <BBB> sancode: don't worry, I won't sue you, I'm just quoting a small part of the license, feel free to ignore ;)
[20:27:57] <ohsix> thats if you use in part the library in an assemblage, and only for one of the possible licenses, as dark_shikari said as a binary and with what they're doing already its kind of moot to pick which one to adhere to
[20:27:58] <sancode> okey
[20:28:16] <BBB> ohsix: but clearly, nothing states that you HAVE to advertise ffmpeg in your about dialog, you could also do it in that startupscreen while the application loads, or in the help, or whatever
[20:28:28] <BBB> ohsix: we just suggest doing it in the about dialog, since that's where you'd expect to find it
[20:28:48] <ohsix> i get the suggestion part, and the niceties
[20:29:38] <ohsix> but its basically an idle threat as its written and ever mentioned ;]
[20:30:25] <sancode> main goal of application is download manager
[20:30:41] <sancode> ffmpeg like extension, program will be work without ffmpeg
[20:30:56] <sancode> next release we plan to do in installer page for selecting components
[20:31:09] <sancode> user can replace ffmpeg.exe with own
[20:31:13] <sancode> and it will be work too
[20:33:26] <sancode> and we will and links to FFmpeg on downloading page
[20:34:27] <BBB> sancode: sounds like you're doing well, license-wise
[20:34:55] <ohsix> going above and beyond, even; but don't threaten him :\
[20:35:53] <CIA-92> ffmpeg: siretart * r22488 /branches/0.5/libswscale/swscale.c:
[20:35:53] <CIA-92> ffmpeg: fix compilation issue on powerpc
[20:35:53] <CIA-92> ffmpeg: unlike the ARCH_ macros, COMPILE_ALTIVEC needs to be tested more carefully
[20:36:20] <sancode> ohsix, i did not understand your last sentense.. i am not native english speaker (
[20:36:47] <BBB> sancode: he says you're compliant and quite comfortably so
[20:36:50] <ohsix> sancode: its nothing you need to worry about, you're doing fine
[20:37:27] <sancode> ok) i understood now
[20:39:40] <BBB> ohsix: nothing wrong with interpreting clauses of the lgpl in a way that is convenient to yourself
[20:40:13] <ohsix> heh, its not even that, and thats being deceitful anyways
[20:42:26] <sancode> so, recommendations:
[20:42:26] <sancode> 1. add link to FFmpeg on main page/download page
[20:42:26] <sancode> 2. More informative About
[20:42:26] <sancode> 3. "this product contains ffmpeg <link> under the license <license + link>"
[20:42:26] <sancode> 4. Links on main page: source codes/build instructions
[20:42:48] <BBB> yes
[20:43:10] <BBB> and add a link to ffmpeg.org in your EULA
[20:43:16] <BBB> probably flvstreamer also
[20:43:26] <sancode> okey
[20:43:51] <sancode> thanks for your help!
[20:47:45] <sancode> and one question..
[20:48:30] <sancode> references to the licenses should points to our site or gnu?
[20:48:42] <justlooking_> sancode, you are aware your app http://flashvideodownloader.org/flash_video_downloader-1.9.8-fx.xpi says its not compatable with firefox 3.6 right ?
[20:48:42] <sancode> or both?
[20:48:55] <sancode> i don't know i am not developing it
[20:49:29] <sancode> i am working only FVD Suite
[20:49:34] <sancode> *with
[20:53:21] <BBB> sancode: gnu is fine
[20:53:41] <BBB> sancode: so http://www.gnu.org/licenses/old-licenses/lgpl-2.1.html
[20:54:31] <sancode> ok
[21:01:25] <ohsix> Dark_Shikari: you there?
[21:01:54] <ohsix> Dark_Shikari: param to decode_nal_units looks pretty suspect: http://paste.ubuntu.com/394242/
[21:02:35] <ohsix> wont be back for a bit to discuss it, but if you have any tips msg me; i'll be digging in a bit later, thanks!
[21:16:51] <Yuvi> I seem to be asleep most of the time when diego is on irc
[21:34:44] <gangil> merbanan: hi
[21:37:14] <astrange> 12:00 <@Dark_Shikari> astrange has to keep updating his fork <- i'm working on ffmpeg frontend issues now so i can submit it with theora support soon
[21:37:38] <astrange> in case anyone rewrite the vp3 decoder over again before then
[21:38:04] <peloverde> really? you are going to submit it soon?
[21:38:18] <astrange> not h264, but the framework part
[21:38:30] <peloverde> still that's huge
[21:54:59] <siretart`> Yuvi: I'm interested about the status of your ogg and theora work as well :-)
[21:55:03] <astrange> i wanted to do pts "properly" in ffmpeg.c using reordered_opaque, but some demuxers (notably .h264) don't set it right in the first place
[21:55:22] <astrange> so it looks like it's back to semi-hacked tracking decoder delay
[21:56:13] <peloverde> siretart`: What's the status for 0.6? Are we still planning on branching tomorrow?
[21:57:17] <siretart`> peloverde: that's what I'm currently trying to find out :-)
[21:57:44] <siretart`> peloverde: do you feel we're ready yet?
[21:58:21] <peloverde> siretart`: SBR is stable with no known issues. I'm waiting for google security review but don't have a time table on that
[21:59:34] <astrange> i hope i haven't killed that icc thread by offering to diagnose that bug
[21:59:41] <peloverde> I would like to see the regtest-mxf fixed on gcc-svn
[22:00:05] <astrange> what arch/configure line?
[22:00:31] <peloverde> http://fate.multimedia.cx/index.php?build_record=199882
[22:02:22] <astrange> ah, that one. let's see if i can reproduce it this time
[22:03:00] <siretart`> peloverde: in any case, I'll try to discuss creating the branch with diego tomorrow.
[22:03:19] <peloverde> siretart`: OK, sounds good
[22:03:52] <siretart`> as much I'm really looking forward to -mt, AFAIUI its regression potential is pretty high, so releasing before merging it seems reasonable to me
[22:05:48] <astrange> to avoid breaking vlc i'm going to turn it off and require the client to set thread_type appropriately, but i still agree
[22:08:55] <peloverde> For 0.7 I want to see AAC in fate in some form
[22:10:23] <Yuvi> siretart`: oh shit, that's tomorrow?
[22:11:04] <Yuvi> well, the two biggest things I want to finish before 0.6 are 1. 4:2:2/4:4:4 and 2. fixing a bunch of smaller ogg seek bugs
[22:12:02] <peloverde> Yuvi: What's the time table on that?
[22:12:14] <Yuvi> I'll see if I can finish 4:2:2/4:4:4 today
[22:12:45] <Yuvi> the seek stuff, I want to know what issues diego's having with it
[22:13:00] <Yuvi> fixing the worst that I know about can probably be done in another day
[22:13:14] <astrange> http://permalink.gmane.org/gmane.comp.video.ffmpeg.cvs/28612 this needs to be applied
[22:13:20] <astrange> unless it magically compiles for someone else?
[22:17:37] <astrange> 'configure --cc=/usr/local/gcc45/bin/gcc --enable-shared --extra-cflags=-m32 --extra-ldflags=-m32' passes make test
[22:18:01] <astrange> on darwin, trying linux
[22:19:47] <siretart`> Yuvi: is the work you're intending to do "contained"? Read: do you need to introduce new or remove existing files? Or do you expect they can be easily backported?
[22:20:15] <siretart`> Yuvi: if not, waiting another week shouldn't really hurt either
[22:21:14] <Yuvi> it's contained, but the seeking stuff won't be easily backported without porting it all
[22:26:26] <siretart`> okay, then I'll try to talk to diego about the "seek stuff".
[22:27:18] <siretart`> if its a non-issue, I think we could indeed branch tomorrow, and backport the 'remaining' ogg commits
[22:42:57] <Yuvi> astrange: if mru doesn't respond, I say just apply it
[22:55:53] <astrange> LDlibavformat/libavformat.so.52
[22:55:54] <astrange> /usr/bin/ld: internal error in override_version, at ../../gold/resolve.cc:61
[22:55:57] <astrange> collect2: ld returned 1 exit status
[22:56:02] <astrange> better switch back to regular ld
[23:12:16] <astrange> http://pastebin.com/QKCCmZU5
[23:25:42] <CIA-92> ffmpeg: aurel * r22489 /trunk/libavcodec/libxvidff.c: libxvid is producing standard mpeg4 so it don't need its own codec id
[23:26:54] <astrange> if(avctx->rc_buffer_size && avctx->bit_rate*av_q2d(avctx->time_base) > avctx->rc_buffer_size)
[23:27:03] <astrange> this is numerically suspicious
[23:27:15] <astrange> avctx->bit_rate*av_q2d(avctx->time_base) should == avctx->rc_buffer_size
[23:27:21] <astrange> so if it rounds wrong it fails
[23:35:23] <CIA-92> ffmpeg: aurel * r22490 /trunk/libavformat/ (rmdec.c rm.h): export sipr_subpk_size for future use by matroska demuxer
[23:36:48] <CIA-92> ffmpeg: aurel * r22491 /trunk/libavformat/ (rmdec.c rm.h): export rm_reorder_sipr_data() for future use by matroska demuxer
[23:49:55] <CIA-92> ffmpeg: aurel * r22492 /trunk/libavformat/ (Makefile matroskadec.c matroska.c): matroskadec: add support for SIPR audio tracks
More information about the FFmpeg-devel-irc