[Ffmpeg-devel-irc] ffmpeg-devel.log.20120911

[burek]

[00:09] <michaelni> pcbb, jpeg over RTP is recent make sure you have the latest ffmpeg also if it doesnt work please provide more information, full output from ffmpeg and a reproduceable testcase
[00:11] <pcbb> michaelni Yah ... my ffmpeg is updated... well the RTP JPEG is a little different. http://tools.ietf.org/html/rfc2035      "...A special header is added to each packet that immediately follows the RTP header..."
[00:12] <pcbb> I want know if FFMPEG can parse this ...
[00:12] <pcbb> and give me a full jpeg
[00:19] <michaelni> pcbb, rfc2035 / 2435 should be supported
[00:23] <pcbb> michaelni I can't find any topic about my question that is "how convert the AVPacket received from jpeg rtsp to jpeg file"
[00:24] <michaelni> pcbb, the demuxer should provide you with a valid jpeg
[00:25] <michaelni> pcbb, also if you want to debug this yourself, maybe look at libavformat/rtpdec_jpeg.c
[00:25] <michaelni> and the jpeg decoder to see where it fails ...
[00:37] <ubitux> http://cubieboard.org/  "NEON"
[00:38] <ubitux> might be interesting to bookmark for later
[00:39] <pcbb> michaelni
[00:39] <pcbb> what is the difference of muxer vs demuxer?
[00:39] <ubitux> respectively write packets and read packets
[00:41] <pcbb> COOL
[00:41] <pcbb> I've the AVInputFormat... how configure the demuxer?? I can't see this in the documentation
[00:41] <ubitux> look at the doc/examples
[00:42] <pcbb> ok
[00:42] <pcbb> I'll see
[00:42] <ubitux> btw, this is the development channel, not the user/api support channel
[00:44] <pcbb> sorry
[01:02] <pcbb> ubitux don't have examples about demuxers and don't have many informations on documentation
[01:02] <pcbb> -.-
[01:15] <ubitux> pcbb: doc/examples/demuxing.c
[01:17] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * rf740c1a9c2 10ffmpeg/ffprobe.c: 
[01:17] <CIA-56> ffmpeg: ffprobe: add and use DEFINE_WRITER_CLASS macro
[01:17] <CIA-56> ffmpeg: Factorize writer class definition.
[01:17] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r0c71d5a077 10ffmpeg/ (doc/ffprobe.texi ffprobe.c): 
[01:17] <CIA-56> ffmpeg: ffprobe: add print_section option to the compact writer
[01:17] <CIA-56> ffmpeg: Allow to skip section name at the begin of each line. Possibly simplify
[01:17] <CIA-56> ffmpeg: output.
[01:17] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * rac006d29d8 10ffmpeg/libavcodec/flvdec.c: lavc/flvdec: return meaningful error codes
[01:18] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r7b62d3415e 10ffmpeg/libavcodec/h263dec.c: lavc/h263dec: return or propagate proper error codes
[01:29] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r00d516454c 10ffmpeg/ (doc/muxers.texi libavformat/segment.c libavformat/version.h): 
[01:29] <CIA-56> ffmpeg: lavf/segment: add segment_list_flags option
[01:29] <CIA-56> ffmpeg: Allow to specify options affecting the segment list generation.
[01:29] <CIA-56> ffmpeg: In particular: add +live and +cache flags.
[01:29] <CIA-56> ffmpeg: For a full discussion read trac ticket #1642:
[01:29] <CIA-56> ffmpeg: http://ffmpeg.org/trac/ffmpeg/ticket/1642
[01:29] <CIA-56> ffmpeg: Also add live M3U8 generation example.
[01:40] <ubitux> argl again i'm sending to mail in private
[01:40] <ubitux> !@#
[02:12] <ubitux> saste: hey btw, shouldn't we think of adding -af to ffplay at some point? :P)
[02:16] <ubitux> :(
[03:26] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r386d60f978 10ffmpeg/libavcodec/aacdec.c: 
[03:26] <CIA-56> ffmpeg: aacdec/lcg_random: fix harmless integer overflow
[03:26] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:26] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r8470e6c96f 10ffmpeg/libavformat/utils.c: 
[03:26] <CIA-56> ffmpeg: lavf/avformat_seek_file: fix harmless integer overflow
[03:26] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:26] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * ref912134f0 10ffmpeg/libswscale/x86/swscale.c: 
[03:26] <CIA-56> ffmpeg: x86/swscale: use unsigned to build SIMD values to avoid signed overflow
[03:26] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:27] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r7ea5dbcb83 10ffmpeg/libavcodec/cavsdec.c: 
[03:27] <CIA-56> ffmpeg: cavsdec: Fix assertion failure.
[03:27] <CIA-56> ffmpeg: The assert in decode_mb_b() is replaced by a normal error check for the
[03:27] <CIA-56> ffmpeg: case that can occur with a damaged or crafted bitstream.
[03:27] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:30] <oneal> Hello, I want to know are there any body who are focusing on parallizing the video transcode pipeline. eg, the decode stage, filter processing stage and encode stage are proccessing in a different threads. 
[03:32] <Compn> theres some people
[03:32] <Compn> decode and encode are both threaded 
[03:32] <Compn> i dont know if filters are being threaded yet
[03:32] <oneal> Yes, but the filter stages is not
[03:33] <Compn> have to talk to saste, see if the api is there yet
[03:33] <Compn> i think saste is still working on porting filters atm.
[03:33] <Compn> ubitux may also know
[03:33] <oneal> who is focusing on this. are there some patchs submitted .
[03:35] <oneal> Compn, you mean ubitux and saste know about this.
[03:36] <Compn> i mean saste is the filter guy
[03:36] <Compn> so he knows the api
[03:36] <Compn> saste = stefano 
[03:36] <Compn> if you want to mail him...
[03:37] <oneal> Thanks, Compn, I want to talk about this with him
[03:38] <Compn> are you interested in contributing ?
[03:38] <ubitux> i'm having fun with lavfi, but nothing about //
[03:38] <Compn> or just really want that feature?
[03:39] <oneal> Yes,I'm doing some optimizations on ffmpeg, 
[03:39] <oneal> accelerating the filter processing and the pipeline.
[03:43] <oneal> the current case in ffmpeg is the current frame is proccessed by the filter, then this frame will be encoded, after encoding the current frame, then the filter only process the next frame.  
[03:49] <ubitux> http://ubitux.fr/pub/pics/ebur128.png \o/
[03:49] <ubitux> let's add some fancy info now
[04:02] <oneal> ubitux, I don't catch what's your mean.
[04:05] <ubitux> i'm just talking for myself :p
[04:05] <oneal> hmm, I see. 
[05:04] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * rbb4ddeea5f 10ffmpeg/libavutil/xtea.c: 
[05:04] <CIA-56> ffmpeg: xtea: optimize core, about 4-5% faster.
[05:04] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:45] <ubitux> if i want to know if a given channel layout has a LFE and get the id of the channel, how am i supposed to do that?
[05:46] <ubitux> i'm checking (channel_layout & 1ULL<<i) == AV_CH_LOW_FREQUENCY right now (for each channel i)
[05:46] <ubitux> is the LFE always supposed to be at the same id?
[07:17] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * rba9e9605ad 10ffmpeg/libavcodec/cavsdec.c: 
[07:17] <CIA-56> ffmpeg: cavsdec/decode_mb_b: fix return type
[07:17] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[07:45] <ubitux> http://lucy.pkh.me/loudness.webm :)
[07:57] <ubitux> i need to add print text, but i'm unsure about how to achieve that
[07:57] <ubitux> i don't like very much adding a fontconfig dependency
[07:57] <ubitux> since it's basically to print numbers and a few ascii letters
[07:58] <ubitux> it sounds almost better to me to use a bintext encoder
[08:54] <ubitux> michaelni: decode_mb_b() returns void
[08:54] <ubitux> so your last cavsdec doesn't look ok
[08:55] <ubitux> +patch
[12:19] <CIA-56> ffmpeg: 03Carl Eugen Hoyos 07master * r20361bcc26 10ffmpeg/configure: The TrueHD decoder needs the MLP parser.
[13:24] <michaelni> ubitux,  theres a return AVERROR_INVALIDDATA in decode_mb_b() so the chnage to int was needed
[14:54] <CIA-56> ffmpeg: 03Clément BSsch 07master * r7b56dddd56 10ffmpeg/libavfilter/af_amerge.c: 
[14:54] <CIA-56> ffmpeg: lavfi/amerge: fix input pad name behaviour.
[14:54] <CIA-56> ffmpeg: This fixes two problems:
[14:54] <CIA-56> ffmpeg:  - the invalid pad.name stack pointer after init() ends
[14:54] <CIA-56> ffmpeg:  - the duplicated name for all inputs
[14:58] <ubitux> michaelni: i didn't see you actually fixed it
[14:59] <ubitux> sorry, i was refering to 7ea5dbcb
[14:59] <ubitux> but you fixed it quickly
[15:02] <saste> ubitux: can you compare amerge with amix?
[15:02] <saste> what are the main differences?
[15:02] <saste> at least i see that amix has some options which we lack in amerge
[15:03] <ubitux> mmmh
[15:03] <saste> getting the longest stream it is a nice feature indeed
[15:03] <ubitux> no idea
[15:03] <ubitux> i think Nicolas really doesn't like amix for several reasons
[15:03] <ubitux> but i don't remember them very well
[15:34] <CIA-56> ffmpeg: 03Martin Storsjö 07master * r88a3569917 10ffmpeg/configure: (log message trimmed)
[15:34] <CIA-56> ffmpeg: configure: Don't try to enable the log2 function on msvcrt
[15:34] <CIA-56> ffmpeg: Some msvcrt versions (the static 64 bit libc in MSVC 10) have
[15:34] <CIA-56> ffmpeg: a log2 function, but there is no declaration for it in the headers.
[15:34] <CIA-56> ffmpeg: Therefore, the normal configure check might find it, but it can fail
[15:34] <CIA-56> ffmpeg: during build or at runtime, depending on whether implicit function
[15:34] <CIA-56> ffmpeg: declarations are an error or not.
[15:34] <CIA-56> ffmpeg: 03Diego Biurrun 07master * rafb5ed55d2 10ffmpeg/configure: 
[15:34] <CIA-56> ffmpeg: configure: Add --disable-inline-asm command line option
[15:34] <CIA-56> ffmpeg: This can come in handy for testing and possibly other purposes.
[15:34] <CIA-56> ffmpeg: 03Martin Storsjö 07master * r0697d81269 10ffmpeg/libavformat/file.c: 
[15:34] <CIA-56> ffmpeg: file: Use a normal private context for storing the file descriptor
[15:34] <CIA-56> ffmpeg: Previously the file descriptor was stored in the priv_data pointer.
[15:34] <CIA-56> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:35] <CIA-56> ffmpeg: 03Anton Khirnov 07master * r0e8dccb79b 10ffmpeg/avplay.c: avplay: get rid of ugly casts in the options table
[15:35] <CIA-56> ffmpeg: 03Diego Biurrun 07master * r1648a508fa 10ffmpeg/libavcodec/x86/dsputil_mmx.c: 
[15:35] <CIA-56> ffmpeg: x86: dsputil: Move specific optimization settings out of global init function
[15:35] <CIA-56> ffmpeg: They belong in the init functions specific to each CPU capability.
[15:35] <CIA-56> ffmpeg: 03Anton Khirnov 07master * ree0dadc109 10ffmpeg/libavformat/flvdec.c: 
[15:35] <CIA-56> ffmpeg: flvdec: always set AVFMTCTX_NOHEADER.
[15:35] <CIA-56> ffmpeg: New streams may be created at any time, e.g. on codec change.
[15:35] <CIA-56> ffmpeg: 03Martin Storsjö 07master * r5fa22ae346 10ffmpeg/configure: (log message trimmed)
[15:35] <CIA-56> ffmpeg: configure: Adjust the xgetbv instrinsic check
[15:35] <CIA-56> ffmpeg: The 64 bit cl.exe version 16.00.30319.01 crashes with an internal
[15:35] <CIA-56> ffmpeg: compiler error on the current check (and thus deduces it isn't
[15:35] <CIA-56> ffmpeg: supported, even if the actual usage in libavuil/x86/cpu.c works
[15:35] <CIA-56> ffmpeg: fine), but by assigning the value from the intrinsic to a variable,
[15:35] <CIA-56> ffmpeg: or returning it, it works fine.
[16:28] <saste> michaelni: 75a9479b393a02ffa03436f275ad2b91bd2acd26
[16:28] <saste> can you comment on it?
[16:28] <ubitux> saste: did you get my PV?
[16:28] <saste> i was trying to write a rawvideo AVI and the generated file fails to be sanely recognized
[16:41] <michaelni> saste, i need to know more details to fix it ...
[16:45] <michaelni> Retransmit: saste, i need to know more details to fix it ...
[16:46] <saste> michaelni, try this command
[16:47] <saste> michaelni, ffmpeg -f lavfi -i testsrc=d=5 -codec:v rawvideo -pix_fmt rgb24  -y out.avi
[16:47] <saste> then ffprobe/ffplay/vlc out.avi
[16:48] <Daemon404> question: why does mt-work/ still exist?
[16:48] <saste> avformat_write_header() sets codec_tag to 0, then the AVI demuxer picks bgr24 when guessing the pixel format
[17:17] <Daemon404> also, in here, nevcairiel, ffmpeg only needs a few more patches to work with msvc
[17:17] <Daemon404> one is annoying though.
[17:19] <nevcairiel> whats the annoying one?
[17:19] <Daemon404> for the host-tool print-options
[17:19] <Daemon404> its a makefile and/or configure thing
[17:19] <Daemon404> i think
[17:37] <saste> how can i check which headers are missing in a .c file?
[17:37] <saste> the equivalent of make checkheaders
[17:39] <Daemon404> is andoma still around?
[17:39] <Daemon404> maybe astrange knows
[17:40] <saste> Daemon404: andoma, astrange
[17:40] <saste> can't say them here since ages
[17:40] <saste> *can't see
[17:41] <Daemon404> saste, i want to send an RFC to the ML to rm -r mt-work/, and i thought about moving yuvcmp.c to tools/
[17:41] <Daemon404> but it has no license
[17:41] <Daemon404> just "written by <name>"
[17:41] <Daemon404>  * originally by Andreas Öman (andoma)
[17:41] <Daemon404>  * some changes by Alexander Strange
[17:41] <saste> ah ok
[17:41] <saste> send the patch and keep the no-license
[17:42] <Daemon404> it's just an RFC
[17:42] <Daemon404> and comments we shall get
[17:42] <saste> you may CC one or both of them
[17:42] <Daemon404> mt-work is dead code otherwise
[17:48] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r2fe8fd3948 10ffmpeg/doc/examples/demuxing.c: examples/demuxing: add audio decoding/demuxing
[17:51] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * ra990a30883 10ffmpeg/libavcodec/ (mpegvideo.h mpegvideo_enc.c ratecontrol.c): 
[17:51] <CIA-56> ffmpeg: ratecontrol: correct predictor in case of stuffing
[17:51] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:51] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * rce7876cbf4 10ffmpeg/libavcodec/ (options_table.h ratecontrol.c): 
[17:51] <CIA-56> ffmpeg: ratecontrol: Dynamically choose a default for rc_max_available_vbv_use
[17:51] <CIA-56> ffmpeg: This improves handling of small buffer sizes
[17:51] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:51] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r50dba71b7d 10ffmpeg/libavcodec/ratecontrol.c: 
[17:51] <CIA-56> ffmpeg: ratecontrol: fix get_fps() argument
[17:51] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:51] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r0ea65bbeee 10ffmpeg/libavcodec/ratecontrol.c: 
[17:51] <CIA-56> ffmpeg: ratecontrol: change a commented printf to av_log
[17:51] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:55] <michaelni> status.github.com "Major service outage" :(
[17:55] <Daemon404> inb4 major loss
[17:56] <Daemon404> i dont trust people who write a website in ruby and erlang that runs in The Cloud to back stuff up at all
[17:56] <Daemon404> <_<
[18:09] <gnafu> Anyone who has their code on [any one thing] and nowhere else is...  Well, I mean, really.  Do I even have to say it?
[18:28] <Daemon404> oh noes, more msvc breakage... from today!
[18:28] Action: Daemon404 slaps ubitux 
[18:28] <ubitux> did i break something?
[18:28] <Daemon404> hmm maybe not
[18:28] <Daemon404> it suddenly fails on af_merge.c
[18:29] <ubitux> oh?
[18:29] <Daemon404> i just blamed the last person to touch it
[18:29] <ubitux> that's possible it's me then
[18:29] <ubitux> what's the failure?
[18:29] <Daemon404> line 314
[18:29] <Daemon404> illegal use of avfilterpad as an expression
[18:29] <ubitux> oh
[18:30] <ubitux> just a sec
[18:32] <Daemon404> ubitux, oh
[18:32] <Daemon404> it looks like its just bitching about mixed code and var decls
[18:32] <Daemon404> easy fix
[18:32] <CIA-56> ffmpeg: 03Clément BSsch 07master * rbbae8cdf4d 10ffmpeg/libavfilter/af_amerge.c: lavfi/amerge: avoid a forward declaration.
[18:32] <Daemon404> you can send a patch or i can
[18:32] <ubitux> here you go
[18:32] <Daemon404> oh
[18:32] <Daemon404> well that was fats.
[18:32] <Daemon404> fast even
[18:33] <ubitux> is it possible to make your msvc instance appears on FATE?
[18:33] <Daemon404> ubitux, ffmpeg doesnt build on fate yet
[18:34] <Daemon404> so theyre only on libav's fate
[18:34] <Daemon404> im working on it
[18:34] <Daemon404> notice my many patches :P
[18:34] <ubitux> it's not possible to make it build with a few flags temporarly?
[18:34] <ubitux> like --disable-pp and such
[18:34] <Daemon404> no.
[18:34] <ubitux> ok :(
[18:34] <Daemon404> ffmpeg.c needs a patch
[18:34] <Daemon404> comign soon
[18:34] <Daemon404> so does print-options need one
[18:34] <Daemon404> (to configure)
[18:35] <ubitux> ok
[18:35] <ubitux> thanks for your work :)
[18:35] <ubitux> r
[18:35] <ubitux> oups
[19:03] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r3015194448 10ffmpeg/libavfilter/af_amerge.c: lavfi/amerge: clarify the error message in case of input overlap
[19:03] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * re592fd0ed1 10ffmpeg/libavformat/flvenc.c: 
[19:03] <CIA-56> ffmpeg: lavf/flvenc: convert commented debug log to an av_dlog call
[19:03] <CIA-56> ffmpeg: Also add some additional info (DTS and DTS time).
[19:25] <Daemon404> hmm
[19:25] <Daemon404> i wish ffmpeg had patchwork
[19:25] <Daemon404> such a pain to save my mails and send them to a diff comp
[19:27] <ubitux> http://patches.libav.org/project/FFmpeg-devel/list/ ?
[19:28] <nevcairiel> lol
[19:28] <ubitux> :D
[19:28] <ubitux> dunno how that work though
[19:28] <nevcairiel> seems its broken
[19:28] <nevcairiel> no patches since 2011
[19:28] <Daemon404> wat
[19:28] <nevcairiel> must've turned it off for ffmpeg-devel
[19:28] <ubitux> ah, yeah
[19:28] <Daemon404> oh that
[19:28] <Daemon404> its old yes
[19:28] <nevcairiel> i mean, it is the libav patchwork
[19:28] <Daemon404> i was gonna set up patchwork
[19:28] <Daemon404> but... i need django and shit
[19:29] <Daemon404> it looks like pain
[19:29] <ubitux> :)
[19:29] <ubitux> looks like they keep debug=True :p
[19:48] Action: ubitux scratches j-b for a rtmp window vm
[19:56] <Daemon404> oh!
[19:56] <Daemon404> print-options suddenly works
[19:56] Action: Daemon404 wonders what happened
[19:58] <Daemon404> anyone very versed in va_list stuff? (this is for cmdutils.c)
[19:58] <Daemon404> i.e. msvc does not have va_copy
[19:59] <ubitux> #define va_copy(a,b) a=b ?
[19:59] <Daemon404> thats not portable
[19:59] <Daemon404> to anything but msvc
[19:59] <ubitux> well, just for msvc
[19:59] <Daemon404> seems rather ugly
[19:59] <ubitux> don't we have a place for all that stuff?
[20:00] <Daemon404> iirc va_copy is nto required
[20:00] <Daemon404> to write portable code
[20:00] <Daemon404> id rather just get rid of it if possible
[20:00] <ohsix> all it does is copy the initial pointer, just like a=b
[20:00] <Daemon404> ohsix, thats not true for all systems
[20:01] <ohsix> fo course, if the stack layout is interesting it can't be done with just a pointer
[20:01] <ubitux> Daemon404: va_copy is a c99 feature, why not make the c99toc89 handle that simple case?
[20:02] <Daemon404> is it -actually- c99?
[20:02] <Daemon404> as in the spec?
[20:02] <ubitux> according to the man :p
[20:02] <Daemon404> looks like it
[20:02] <ubitux>        The va_start(), va_arg(), and va_end() macros conform to C89.  C99 defines the va_copy() macro.
[20:02] <Daemon404> hmmm
[20:02] <Daemon404> interesting
[20:02] <Daemon404> might be better...
[20:02] <ohsix> maybe you don't get c99 on interesting stacks? :D
[20:03] <ohsix> there's not much reason to need va_copy, in any situation where you might you could change the code so you don't
[20:03] <ubitux> OTOH, doing this in c99toc89 won't generate valid c89 code on all systems
[20:03] <Daemon404> yeah
[20:03] <Daemon404> libav is actually running a converter+gcc instance
[20:03] <Daemon404> to test such things
[20:03] <ubitux> still, it looks like it belongs there
[20:04] <ubitux> we use va_copy in bprint api too
[20:04] <Daemon404> ohsix, http://www.google.com/#hl=en&sclient=psy-ab&q=va_copy+c99&oq=va_copy+c99&gs_l=serp.3..0j0i10i30j0i30l2.55346.55777.0.55991.4.4.0.0.0.0.537.818.2-1j5-1.2.0.les%3B..0.0...1c.1.xG90SnVxiZc&pbx=1&bav=on.2,or.r_gc.r_pw.r_qf.&fp=d8007f785e356eff&biw=1920&bih=1045
[20:04] <Daemon404> woops
[20:04] <Daemon404> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=cmdutils.c;h=a0ad3ef747d40a67ac04ada87ff01aef07f99d18;hb=HEAD#l103
[20:04] <Daemon404> ubitux, not anymore
[20:04] <Daemon404> or, soon
[20:04] <ohsix> zomg that had your in real life earth geolocation in the url
[20:05] <ubitux> ok
[20:05] <Daemon404> /care
[20:05] <Daemon404> its my workplace
[20:05] <Daemon404> everyone knows where i work
[20:06] <Daemon404> ffmpeg likely needs around 4 patche to be msvc compatible
[20:06] <Daemon404> postproc fix, which is in v2 and awaiting michaelni 
[20:06] <ohsix> va_end with no start? also you only need to copy the va if you want to repeatedly do va_arg, then do it again
[20:06] <ohsix> so if there are 3 arguments, va_arg 1, 2, or 3; then use the copy to reset the sequence
[20:06] <Daemon404> ohsix, yeah...
[20:07] <Daemon404> im thinking it isnt even needed
[20:07] <Daemon404> so fixes needed: cmdutils, ffmpeg.c, postproc, bprint
[20:07] <ohsix> invoking a function with va_list isn't going to change it's value :D
[20:07] <Daemon404> ohsix, should i just be able to use va_start
[20:07] <Daemon404> shouldnt*
[20:07] <ohsix> i don't think you even need that, va_start is to set up va_arg
[20:08] <ohsix> i think va_start just assures it's not in a register
[20:08] <Daemon404> then wtf is ca_copy even doing
[20:08] <ohsix> but a) that code looks wrong anyways b) probably don't need to even touch it
[20:08] <Daemon404> s/ca/va/
[20:09] <ohsix> copying a pointer
[20:09] <Daemon404> but why
[20:09] <ohsix> think of the state in strstr, it's sort of the same thing
[20:09] <Daemon404> fun...
[20:09] <ohsix> va_arg just advances the pointer and returns what it pointed at
[20:09] <Daemon404> we arent modifying it anyway
[20:09] <ohsix> so if you're going to start again, you need the starting point
[20:09] <Daemon404> so i dont see why va_copy is needed at all
[20:10] <ohsix> right, it's not; but i'd check the specifications just the same, it's very weird to call va_copy then va_end, so much that i'd be looking at the specs anyways ;]
[20:10] <Daemon404> or i could just ask nicolas
[20:10] <Daemon404> who wrote it
[20:11] <ohsix> the man page says each invocation of va_copy must be matched by a va_end
[20:11] <Daemon404> it is
[20:11] <ohsix> right
[20:11] <Daemon404> also
[20:11] <Daemon404> i just notice
[20:11] <Daemon404> it uses fputs
[20:11] <Daemon404> how evil
[20:12] <ohsix> interesting, va_start can allocate memory to move the arguments out of registers on some systems too
[20:13] <ohsix> so you need the va_copy if you are going to pass va_list to another context :<
[20:13] <ohsix> i've got some code to fix that only coincidentally works at the moment :>
[20:13] <Daemon404> ;p
[20:14] Action: Daemon404 wishes nicolas used irc
[20:16] <CIA-56> ffmpeg: 03Derek Buitenhuis 07master * re20ac54f3e 10ffmpeg/libpostproc/ (postprocess.c postprocess_template.c): 
[20:16] <CIA-56> ffmpeg: postproc: Fix unprotected inline asm
[20:16] <CIA-56> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[20:16] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:17] <ubitux> Daemon404: me too :(
[20:19] <Daemon404> ubitux, i sent him an email
[20:19] <Daemon404> o shit
[20:19] <Daemon404> postproc went it
[20:19] <Daemon404> wizards.jpg
[20:20] <brimestone> hey guys, what causes my ffmpeg to error out when stiching sequencial images to video...
[20:20] <brimestone> http://pastebin.com/yhnnPYkX
[20:20] <ubitux> -f image2 -i foo%d.jpg ?
[20:21] <ubitux> your ffmpeg is old btw, we don't use svn since& well, prehistoric times
[20:21] <ubitux> (this is a user question)
[20:21] <Compn> brimestone : what are you doing in this developer channel with a user question ?
[20:21] <Compn> get out!
[20:21] <ubitux> haha
[20:21] <brimestone> sorry.. user doesnt know too 
[20:21] <Compn> get out get out get out! dexter is a cookie!
[20:22] <Daemon404> brimestone, uh
[20:22] <ubitux> cookie? *_*
[20:22] <Daemon404> put the filename in single quotes
[20:22] Action: Daemon404 wonders if teh shell is escaping it
[20:23] <ubitux> imo this is just missing the image2 format
[20:23] <ubitux> cause the globbing is likely part of this but well
[20:23] <Daemon404> hehe
[20:28] <brimestone> didn't work
[20:28] <brimestone> im re-compiling the new ffmpeg
[21:03] <michaelni> saste, i see no difference from before 75a9479b393a02ffa03436f275ad2b91bd2acd26 and HEAD both show wrong colors with avi output
[21:04] <saste> michealni: uhm yes i see
[21:04] <saste>  //the current rawvideo encoding system ends up setting the wrong codec_tag for avi/mov, we override it here
[21:04] <saste> does it means that AVI muxing is broken with rawvideo?
[21:05] <saste> basically we're not setting the codec tag, so there is no way the AVI demuxer can guess which is the correct pixel format
[21:09] <michaelni> rgb is normally stored with codec_tag 0 in avi and using a bit per pixel field to identify the kind
[21:11] <saste> michaelni, so do you have an idea why rgb24 is probed like bgr24?
[21:12] <saste> i didn't try with other pixel formats, to say the truth
[21:13] <michaelni> do we have a valid avi sample that contains rgb24 ?
[21:13] <michaelni> if so its probably not hard to implement support for it
[21:35] <saste> michaelni: ffmpeg -f lavfi -i testsrc=d=5 -codec:v rawvideo -pix_fmt rgb24  -y out.avi
[21:35] <saste> if you change -pix_fmt rgb24 with bgr24  it will work fine
[21:36] <saste> so it seems the demuxer cannot distinguish between bgr24 and rgb24
[21:37] <saste> with other formats it is even worse (e.g. rgb4 detected as PAL8)
[21:39] <michaelni> if you have / can find samples that are known to be correctly storing rgb4 & rgb24 then it should be easy to implement them
[21:40] <saste> michaelni, ok
[21:41] <saste> michaelni, can you explain why the codec_tag mechanism is not used/ is considered broken?
[21:44] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * rbdd9bfbbab 10ffmpeg/doc/examples/muxing.c: 
[21:44] <CIA-56> ffmpeg: examples/muxing: prefer AVPicture to AVFrame, when feasible
[21:44] <CIA-56> ffmpeg: Favor the use of plain AVPicture over AVFrame, especially when the use of
[21:44] <CIA-56> ffmpeg: AVFrame is not required like in the case of tmp_picture.
[21:44] <CIA-56> ffmpeg: Also adopt more straightforward names, to avoid frame/picture confusion.
[21:44] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r34b5b735f9 10ffmpeg/doc/examples/muxing.c: examples/muxing: add check on avformat_write_header() result
[21:44] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r976bb42aca 10ffmpeg/doc/examples/muxing.c: examples/muxing: merge some declarations and definitions
[21:44] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r1d0048cfe3 10ffmpeg/doc/examples/muxing.c: 
[21:44] <CIA-56> ffmpeg: examples/muxing: rename img_convert_ctx to sws_ctx
[21:44] <CIA-56> ffmpeg: The new name is more consistent with the codebase, and more
[21:44] <CIA-56> ffmpeg: self-consistent with the libswscale API.
[21:44] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * rb3e2975ab3 10ffmpeg/doc/examples/muxing.c: examples/muxing: apply misc fixes to log messages
[22:09] <Daemon404> boooo
[22:09] <Daemon404> swr fate tests crash ffmpeg
[22:10] <nevcairiel> how many patches missing from git now?
[22:10] <Daemon404> the three i just sent, and im working with nicolas to figure out the 4th
[22:10] <Daemon404> then converter needs to be fixed
[22:10] <Daemon404> cause it trips an assert on ffprobe.c
[22:10] <nevcairiel> heh
[22:10] <Daemon404> and i need to figure out why this fate test crashes ffmpeg.exe
[22:11] <nevcairiel> maybe a similar asm issue as avr has :P
[22:11] <Daemon404> ;p
[22:12] <Daemon404> ill try and build with debug symbols
[22:12] <Daemon404> i dont know how to add it with cl.exe
[22:12] <Daemon404> off the top of my head
[22:13] <Daemon404> (what to add to cflags and ldflags)
[22:14] <nevcairiel> --enable-debug should be set to produce it
[22:14] <Daemon404> oic
[22:14] <Daemon404> works for msvc?
[22:14] <nevcairiel> i saw something in the flag filtering for it
[22:14] <Daemon404> vsynth1 and vsynth2 work with ffmpeg.exe
[22:14] Action: Daemon404 goes back to swr
[22:16] <nevcairiel> might need /debug in ldflags, not sure how it behaves whtn the objects have debug info and the linker is not instructed to keep it explicitly
[22:17] <nevcairiel> i'll try that soon enough either way
[22:17] Action: Daemon404 is doing it now
[22:17] <Daemon404> only 8 more years will configure finishes
[22:18] <nevcairiel> full fate run just took (only) 8 minutes, i wonder how much of that time was configure :p
[22:18] <Daemon404> lol
[22:18] <Daemon404> it takes longer than build for me
[22:18] <nevcairiel> about the same time here, 3 minutes each, 2 minutes for the fate run
[22:18] <nevcairiel> crazy
[22:19] <Daemon404> i dont see yout x64 instance
[22:19] <Daemon404> it isnt listed
[22:19] <nevcairiel> its way down with the other x64 instances
[22:19] <Daemon404> oh it is
[22:19] <Daemon404> yeah
[22:21] <Daemon404> oh. BBB just sent the 4th patch to libav's list
[22:23] <nevcairiel> anyone still running pre-XP deserves to be screwed anyway
[22:23] <nevcairiel> personally, i would include people running XP into that group, but i'll probably be chased by an angry mob
[22:23] <Daemon404> lol
[22:23] <Daemon404> i support whatever ms supports
[22:24] <nevcairiel> XP is still up to 2014
[22:24] <Daemon404> i know
[22:24] <Daemon404> i wish it werent
[22:32] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r749ddc14fc 10ffmpeg/ffprobe.c: ffprobe: add priv_class field to Writer, and factorize writer options initialization
[22:33] <saste> nevcairiel: xp is the only windows i can withstand
[22:36] <Daemon404> nevcairiel, where is your ar wrapper
[22:36] <nevcairiel> lib.exe -nologo $(echo $* | sed 's/rc /-out:/') 
[22:36] <nevcairiel> go nuts
[22:36] <Daemon404> lol
[22:37] <nevcairiel> i simply named it "ar" and put it into path so i dont need a --ar param
[22:37] <nevcairiel> should maybe figure out how to make configure aware of lib.exe
[22:38] <nevcairiel> although after poking configure and the makefiles for an hour or so today to try to figure out shared builds, my head still hurts
[22:39] <Daemon404> there we go
[22:39] <Daemon404> lib.exe is needed for debug builds
[22:39] <Daemon404> otherwise it wont link
[22:39] <nevcairiel> aha
[22:43] <Daemon404> and it works
[22:45] <nevcairiel> got some nice debug info?
[22:45] <Daemon404> indeed
[22:45] <Daemon404> but i need ot get it to michael...
[22:45] <Daemon404> and i have no clue how to export a report
[22:46] <nevcairiel> a nice way is to save a dmp file, but that only works if whoever you're sending it to also has msvc, so ... no :P
[22:46] <michaelni> saste, not sure what you mean by "not used", avi codectags are as they are in avi, its how the thing is designed
[22:46] <Daemon404> nevcairiel, yeah
[22:46] <nevcairiel> but those files are still nice and full of infos
[22:47] <Daemon404> interestingly
[22:47] <Daemon404> it crashes in af_pan.c
[22:47] <nevcairiel> asm?
[22:47] <Daemon404> no.
[22:47] <Daemon404> want a .dmp?
[22:47] <Daemon404> :P
[22:48] <saste> michaelni, yes but they are not set by the muxer, at least from what i can see (rawvideo)
[22:57] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r30a265f058 10ffmpeg/libavdevice/lavfi.c: lavd/lavfi: apply stylistic fixes to the options help fields
[23:00] <CIA-56> ffmpeg: 03Derek Buitenhuis 07master * r578bbf8149 10ffmpeg/libavcodec/x86/diracdsp_mmx.c: 
[23:00] <CIA-56> ffmpeg: diracdsp: Fix unprotected inline asm
[23:00] <CIA-56> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[23:00] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:00] <CIA-56> ffmpeg: 03Derek Buitenhuis 07master * r23a5a24c0f 10ffmpeg/ffmpeg.c: 
[23:00] <CIA-56> ffmpeg: ffmpeg: Only include unistd.h if it exists
[23:00] <CIA-56> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[23:00] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:07] <Daemon404> nevcairiel, https://dl.dropbox.com/u/77349181/ffmpeg_dump.7z
[23:08] <nevcairiel> hm, guess i need the binary that crashes for full debug infos
[23:08] <Daemon404> sec
[23:08] <nevcairiel> would've thought in a 50mb dump file he could sneak that in there
[23:08] <nevcairiel> but noo :p
[23:09] <Daemon404> uploading
[23:09] <Daemon404> dat 512 kbit/s upload
[23:12] <Daemon404> nevcairiel, https://dl.dropbox.com/u/77349181/ffmpeg_bin.7z
[23:13] <nevcairiel> ah nice, with sources and everything
[23:13] <nevcairiel> in a few month, i will not remember how i ever managed without it
[23:13] <Daemon404> lol
[23:14] <Daemon404> nevcairiel, you should know i did:
[23:15] <Daemon404> #define va_copy(a, b) (a = b)
[23:15] <Daemon404> but thats only for logging callback
[23:15] <Daemon404> and shouldnt be an issue here
[23:20] <nevcairiel> the issuer there is odd
[23:20] <nevcairiel> the value of the "len" parameter is very close to the pointer of "arg0"
[23:20] <nevcairiel> somehow it got in there
[23:21] Action: Daemon404 has no idea
[23:30] <nevcairiel> oh well, time for sleep
[23:43] <Daemon404> night
[00:00] --- Wed Sep 12 2012