[Ffmpeg-devel-irc] ffmpeg-devel.log.20121015

burek burek021 at gmail.com
Tue Oct 16 02:05:02 CEST 2012


[00:45] <cone-20> ffmpeg.git 03Michael Niedermayer 07a63d7213b3eb: ffplay: Prevent 0/0 aspect from being passed on to lavfi * 03http://tinyurl.com/8touvp203
[00:45] <cone-20> ffmpeg.git 03Michael Niedermayer 071b8f4d773c4d: ffmpeg: use framerate not fieldrate as filter input. * 03http://tinyurl.com/9lkfa4603
[01:27] <Daemon404> /home/daemon404/build_ffmpeg/src/tests/fate-run.sh: line 68: 28651 Illegal instruction $target_exec $target_path/"$@"
[01:27] <Daemon404> wtf?
[01:27] <Daemon404> is this just me?
[03:04] <barque> Guys, I'm getting the same R and G through libavcodec PIX_FMT_RGB24
[03:04] <Compn> command line or your own program ?
[03:04] <barque> my own
[03:05] <barque> I should try cmd line ... I guess .. ? :P
[03:05] <Compn> makes it much easier to test for us
[03:05] <Compn> :)
[03:06] <barque> true, I'm gonna review the code about 10x just to make sure I'm seeing this right
[03:06] <barque> no one has reported something similar have they?
[03:07] <barque> oh my god
[03:07] <barque> this is NOT a problem of yours, but of mine
[03:07] Action: barque crawls into a dark corner
[03:08] <barque> with all this code I was bound to make a mistake anyway :P
[03:08] <barque> first run is never perfect
[03:44] <Compn> poor canaidans :)
[03:44] <Compn> so polite 
[03:44] <Compn> canadians*
[03:57] <cone-20> ffmpeg.git 03Michael Niedermayer 07151469db3379: j2kenc: remove unneeded dereference * 03http://tinyurl.com/9hppmd603
[03:57] <cone-20> ffmpeg.git 03Michael Niedermayer 0735782bfbca0a: h264: Change asserts to av_asserts in ff_h264_fill_default_ref_list() * 03http://tinyurl.com/9o6wehf03
[07:38] <ubitux> nyuhu: so you like fist review?
[08:11] <ubitux> Daemon404: ping
[08:12] <ubitux> any objection for the webm patch according to Nicolas' comment?
[08:17] <Daemon404> ubitux, confirm with the spec then ok
[08:17] <Daemon404> im sleeping
[08:44] <cone-200> ffmpeg.git 03Stefano Sabatini 07cd559bb49a34: lavc: add xface image decoder and encoder * 03http://tinyurl.com/brguubo03
[08:44] <cone-200> ffmpeg.git 03Stefano Sabatini 07304c37b216d7: tests: add fate-xface test * 03http://tinyurl.com/cf2wowx03
[08:55] <ubitux> Daemon404: ok :)
[08:56] <ubitux> yay xface.
[09:20] <divVerent> ubitux: I am back
[09:20] <divVerent> and I got your messages
[09:22] <cone-200> ffmpeg.git 03Clément BSsch 07b08273c9ca8b: lavf/mkv: avoid negative ts by default. * 03http://tinyurl.com/cn5epy503
[12:19] <TimNich> michaelni: ping
[13:02] <cone-200> ffmpeg.git 03Paul B Mahol 074f5e5a05132b: lavf/txd: cosmetics: fix identation * 03http://tinyurl.com/9ock66803
[13:14] <durandal_1707> This program is based on qt-faststart.c from the ffmpeg projects .... 
[13:28] <michaelni> TimNich, pong
[13:32] <TimNich> michaelni:  quicktime is still not right yet, b8d6455  is better but there are still quite a few places where there are wrong values in the mood atoms. See http://pastebin.com/VxHrYdx9
[13:33] <TimNich> s/mood/ moov/
[13:35] <TimNich> if you do a probe you find an error line saying "Timecode frame rate 0/1 not supported
[13:36] <michaelni> TimNich, which values exactly are wrong ?
[13:36] <michaelni> the diff shows which are different not which are wrong ...
[13:38] <TimNich> The full dump is the good one, and in the diff -- good ++ bad
[13:40] <Compn> you mean removing code is good
[13:40] <Compn> and adding code is bad?
[13:40] Action: Compn trying to parse TimNich's sentence
[13:40] <Compn> remember, english isnt everyone's first language here, so be specific
[13:42] <TimNich> Compn:  If you looked at the diff you would see good lines removed, bad lines added. Appreciate the point about non native English speakers, but in context with the diff in front of one I would have thought it made sense...
[13:42] <funman> i guess he ran diff -u good bad
[13:42] <funman> timescale 12800 is weird
[13:43] <Compn> TimNich : i dont even know what mail you are talking about. i'm just in observer mode here 
[13:43] <michaelni> TimNich, well, you want av sync then good is not good at all
[13:43] <TimNich> Compn: 
[13:43] <Compn> i see the pastebin now tho
[13:44] <TimNich> good in terms of what used to work.. the latest version crahses Quicktime player and Avid...
[13:44] <Compn> so you're specifically talking about mov muxer ?
[13:45] <TimNich> If there is an inherent AV sync issue then thats been around a while&.
[13:45] <michaelni> yes
[13:47] <TimNich> Compn: yes, usual story , two steps forward one step back. Sounds like michaelni  may have unearthed a deeper issue though..
[13:48] <michaelni> TimNich, do you have some clue what is exactly causeing the problem ?
[13:48] <michaelni> i mean does QT fundamentally not support such timebases or is it something else ?
[13:53] <TimNich> From a three way diff I wonder about the  timescale: 12800 etc in the mdhd atom of the timecode track as this was the same in both previous versions which at least play after a fashion in QT player
[13:57] <Compn> michaelni : do we have any samples that look like what mov muxer outputs ?
[13:57] <Compn> e.g. 12800 timebase
[13:57] <Compn> i'll look at who changed what in mov muxer in a minute
[13:57] <Compn> :P
[14:00] <TimNich> Compn: We know exactly which commits introduced the issue. Its a case of what mechanism is at work..
[14:00] <michaelni> It would be interresting to have a working mov sample with fine timebase for the video and a timecode track
[14:00] <ubitux> big buck bunny has a timecode track iirc
[14:00] <ubitux> (set to 00:00:00:00 though)
[14:01] <TimNich> what do you mean by "fine" I can knock up anything quicktime Pro can produce..
[14:02] <michaelni> TimNich, with fine i mean for example a 25fps video that has a timebase smaller than 1/25
[14:02] <michaelni> like a 1/1000 timebase or 1/250 or something 
[14:04] <TimNich> michaelni: Not sure what would have that, I can't see QTPro bothering to set the tiembase smaller than it thinks is needed...
[14:10] <TimNich> michaelni: Just looked at a QTPro file and timebase is 25...
[14:13] <TimNich> However I notice an apparent inconsistency in the bad2 headers. In the last stud atom the frame duration is listed add .000078 secs and the number_of_frames 0, both the others have .04 frame_duration
[14:16] <ubitux> michaelni: Dave Rice has a laaarge collection of a lot of mov/mp4 samples with timecodes
[14:16] <ubitux> you might want to contact him
[14:16] <ubitux> (more than 1TB of samples)
[14:17] <ubitux> he is sometimes on IRC btw
[14:24] <cone-200> ffmpeg.git 03Justin Ruggles 078441909f49d1: lavc: update documentation for AVFrame.extended_data * 03http://tinyurl.com/924p3bv03
[14:25] <cone-200> ffmpeg.git 03Justin Ruggles 07977eb7d567f6: shorten: use planar sample format * 03http://tinyurl.com/9zebe7c03
[14:25] <cone-200> ffmpeg.git 03Justin Ruggles 077ebfe5b44a55: wmadec: use float planar sample format output * 03http://tinyurl.com/9st5ua603
[14:25] <cone-200> ffmpeg.git 03Justin Ruggles 072725ce7c7c34: wmalossless: output in planar sample format * 03http://tinyurl.com/9gt236k03
[14:25] <cone-200> ffmpeg.git 03Justin Ruggles 07f9d732c26448: wmapro: use planar sample format * 03http://tinyurl.com/9vf2xp303
[14:25] <cone-200> ffmpeg.git 03Mans Rullgard 07e98b02de5f92: configure: check for mprotect * 03http://tinyurl.com/9vxu5mk03
[14:25] <cone-200> ffmpeg.git 03Mans Rullgard 0795cd815c3663: swscale: try to use mmap only if available * 03http://tinyurl.com/9geu84403
[14:25] <cone-200> ffmpeg.git 03Michael Niedermayer 07ae237a117acb: Merge remote-tracking branch 'qatar/master' * 03http://tinyurl.com/9ylrthd03
[14:48] <ubitux> kierank: you're aware there is a AVDict for metadata in the AVFrame in ffmpeg right?
[14:49] <ubitux> (see f49ec1b4)
[14:51] <durandal_1707> hee is writing it for libav
[14:51] <cone-200> ffmpeg.git 03Paul B Mahol 07916e40b5b42f: smacker: return more meaningfull error codes * 03http://tinyurl.com/8lfhlpa03
[14:52] <ubitux> durandal_1707: yes that's why i'm just saying it's already available and it might be interesting to keep a few things in common
[14:52] <ubitux> again, that's a feature we have since one year now
[14:54] <durandal_1707> nope, since Jul
[14:54] <ubitux> ah my bad, commit date/author date
[14:55] <kierank> i still am unsure frame accuracy will work with threading
[14:55] <kierank> maybe it's ok with a dictionary
[14:55] <kierank> but with a buffer i got a lot of crashes
[14:56] <ubitux> why would you use a buffer?
[14:56] <kierank> to export closed captions
[14:57] <kierank> and arbitary user data
[14:57] <ubitux> any reason not to have it in packet side data?
[14:57] <kierank> because it needs reordering
[14:57] <ubitux> ok
[14:57] <ubitux> anyway, if you add some !dict metadata, please don't call it "metadata"
[14:58] <ubitux> because dict metadata are quite useful
[14:58] <ubitux> we can deal with it without breaking abi (we can replace the getter), but it's a pain
[14:59] <ubitux> kierank: you're trying to support cc & teletext?
[14:59] <kierank> well i just want arbitrary user data
[14:59] <kierank> exported
[14:59] <ubitux> the problem with a dict is that it works only with strings
[15:00] <kierank> yup
[15:00] <ubitux> unless you put some base64 in it
[15:00] <ubitux> but i suppose you're not that mad
[15:10] <cone-200> ffmpeg.git 03Michael Niedermayer 075bac83dae84a: shorten: fix U8 to be planar too * 03http://tinyurl.com/8qu2ex503
[15:10] <cone-200> ffmpeg.git 03Michael Niedermayer 075c7a62aef59e: movenc: fix tmcd parameters * 03http://tinyurl.com/9j6okep03
[15:12] <michaelni> TimNich, can you retry if qt likes the new movs better ?
[15:12] <TimNich> making it now..
[15:28] <TimNich> michaelni: We're good on QT7 and Avid
[15:28] <TimNich> Thanks
[15:30] <cone-200> ffmpeg.git 03Paul B Mahol 075864fe62efa4: flacdec: replace -1 with AV_SAMPLE_FMT_NONE * 03http://tinyurl.com/98x2df603
[15:30] <cone-200> ffmpeg.git 03Paul B Mahol 07e75357ea7c8c: xbmenc: cosmetics: reindent * 03http://tinyurl.com/9x2bm2703
[15:35] <michaelni> Tjoppen, TimNich, mateo` "avio_wb16(pb, mxf->tc.start + frame); // continuity count" <-- the sum does not fit in 16bit
[15:35] <michaelni> what is the right fix for this ?
[15:36] <michaelni> (testcase is simple make fate-lavf-mxf)
[15:36] <michaelni> mxf->tc.start is 264363
[15:36] <cone-200> ffmpeg.git 03Stefano Sabatini 07e56b3a5ebb12: examples/muxing: fix case inconsistency in message * 03http://tinyurl.com/dxnbogu03
[15:37] <cone-200> ffmpeg.git 03Stefano Sabatini 076133149e0c70: examples/muxing: add missing error checks * 03http://tinyurl.com/d4zmd6903
[15:37] <cone-200> ffmpeg.git 03Stefano Sabatini 0759e31aa2cafb: examples/muxing: provide more information in case of avcodec_open2 failure * 03http://tinyurl.com/cpzzj2n03
[15:38] <Tjoppen> michaelni: not a problem, IIRC
[15:43] <Tjoppen> yep, it's supposed to overflow. it was used for detecting mising content packages during transmission (S326m-2000)
[15:47] <michaelni> Tjoppen, ok if i add a & 0xFFFF ?
[15:54] <Tjoppen> yes
[16:07] <cone-200> ffmpeg.git 03Paul B Mahol 07011f3892ff35: h264: use designated initializers for AVClass * 03http://tinyurl.com/93qaqcv03
[16:10] <saste_> can someone comment on https://ffmpeg.org/trac/ffmpeg/ticket/1801?
[16:10] <saste_> what's the proper way to setup DTS anyway?
[16:12] <ubitux> saste_: wait before pushed the lavc/utils patch
[16:12] <ubitux> i think av_get_sample_fmt_name can return null
[16:13] <ubitux> and some libc have some issues with it
[16:13] <ubitux> (like crashing instead of "(null)" or similar)
[16:13] <saste_> ubitux, we have a lot of that
[16:13] <ubitux> yeah i know :(
[16:14] <ubitux> well whatever
[16:14] <saste_> anyway at that stage sample_fmt should be set to a valid value
[16:14] <ubitux> do as you wish :)
[16:14] <saste_> i can do an av_x_if_null()
[16:19] <durandal_1707> why is avio_wl* so stupid?
[16:20] <durandal_1707> yes not using multiple avio_w8 is not making things much faster but still.....
[16:22] <saste_> this is sick
[16:22] <saste_> ffplay -f lavfi "flite=text='I am left':voice=rms[left]; flite=text='I am right'[right]; [left][right]amerge"
[16:22] <saste_> I can hear "left" or "right" according to what I expect to hear
[16:27] <durandal_1707> now make it speak stuff from av_log
[16:28] <saste_> you don't really want that
[16:28] <pettter> gawd, this is such a hack
[16:29] <Tjoppen> I have a simple patch series that adds parsing field order to the DV decoder. is this of interest?
[16:31] <durandal_1707> i'm interested
[16:32] <Tjoppen> it's a massive one-line patch
[16:32] <TimNich> Tjoppen: Not sure, what issue does it solve?
[16:32] <Tjoppen> plus another one before it to move a piece of code up a few lines :)
[16:32] <Tjoppen> it exposes the field order (FS flag). that's it
[16:33] <TimNich> currently bff is assumed?
[16:33] <Tjoppen> dv encoder might want a similar fix, but I don't really care about that
[16:33] <Tjoppen> yes
[16:33] <Tjoppen> the fix for the encoder would be to put !top_field_first << 6 in the relevant byte
[16:33] <Tjoppen> PC3 of the VSC pack
[16:33] <saste_> who can explain this: http://pastie.org/5062307
[16:34] <saste_> sample_fmt is not the one shown when encoding
[16:34] <TimNich> I'd like to know it had done it (AV_LOG_WARNING)
[16:37] <durandal_1707> saste_: isn't that normal for mp3?
[16:37] <TimNich> Actually I think I would really like field_order to be one of those AV_LOG_INFO bits like frame rate etc that utils spits out at various moments&.
[16:40] <pettter> I have a rather hacky patch for #241 and duplicates
[16:43] <Tjoppen> TimNich: uhm, ffprobe -show_streams
[16:43] <Tjoppen> or show_frames
[16:44] <mateo`> Tjoppen: does your fix deal with the FF/FS value in the VSC pack ?
[16:44] <Tjoppen> I considered FF. not sure what to do if it's zero
[16:44] <mateo`> Tjoppen: i think the encoder should be fixed too if i remember well
[16:45] <Tjoppen> prolly
[16:45] <mateo`> anyway i should do that on the encoder part when i'll get some time
[16:45] <Tjoppen> you know what? I'm going to look into that too. FATE passes with extracting FS
[16:46] <Tjoppen> ugh, AV_FIELD_TB/BT
[16:47] <TimNich> Tjoppen: -show-frames, but thats OK for a prodding an unknown stream. If it was added into the other output on a per file basis then it could be checked fro in the logs of batch jobs without the extra pass of a probe...
[16:47] <Tjoppen> I don't see what this has to do with the dv decoder
[16:47] <Tjoppen> that's a CLI issue
[16:48] <TimNich> It doesn't. But my previous comment got me thinking about my more generalised wish list&.
[16:49] <TimNich> Agree with the AV_FIELD_TB/BT but that's Apple's nomenclature, and confusing it is too...
[16:51] <Tjoppen> mateo`: http://pastebin.com/k2chCe38
[16:51] <Tjoppen> TimNich: nm, I just read it from AVPicture instead
[16:56] <TimNich> Tjoppen:  there are too many places too look...
[17:00] <Tjoppen> TimNich: there appears to be two places. but: the TB/BT stuff don't apply to raw frames
[17:05] <TimNich> Well the current spec mandates the use of the fiel data in that format for uncompressed video, and michaelni  said it should get it from the avctx->field_order
[17:06] <TimNich> AVOicture doesn't actually seem relevant, did you mean AVFrame?
[17:13] <Tjoppen> AVFrame picture;
[17:13] <Tjoppen> right
[17:14] <Tjoppen> isn't AVFrame.top_field_first set from AVCodecContext though?
[17:14] <cone-200> ffmpeg.git 03Paul B Mahol 073f8148911c6e: img2dec: check return value of av_new_packet() * 03http://tinyurl.com/8c2dygy03
[17:17] <TimNich> Tjoppen:  not that I could find. I gripped the whole source tree and the only time AVCodecContext->Field_order is read is either for a stream copy, or to see if the file atom should be written to a mov (and if its not set it isn't)
[17:20] <TimNich> michaelni: suggested that it was up to the demuxer to fill it in, and the decoder to move it to the coded_frame->top_field_first, etc
[17:21] <Tjoppen> no, that does not make sense
[17:21] <Tjoppen> the decoder should fill top_field_first from the value actually stored in the essence, if applicable
[17:22] <Tjoppen> lavc ought to deliver what the essence says, and lavf what the container says
[17:22] <TimNich> Agreed, but in the case of the essence not containing that information, but it being available elsewhere, surely one should look there in the second instance?
[17:23] <Tjoppen> correct
[17:23] <Tjoppen> and vice versa
[17:23] <Tjoppen> or.. I might slightly prefer if a separate library did this
[17:24] <Tjoppen> this is sort of my beef with lavf - it's guessing too much. I'd rather it were dumber, and having something like libavguess instead :)
[17:24] <TimNich> Well if you ffprobe -show_frames a mov containing uncompressed video it will give you a 0 0
[17:24] <TimNich> Even though it is interlaced tff which the file atom will tell you.
[17:25] <Tjoppen> so the raw video decoder needs to fill in top_field_first
[17:25] <TimNich> yes
[17:25] <mateo`> Tjoppen: the patch should be ok :)
[17:25] <Tjoppen> nice. I'll post that and the other two
[17:30] <TimNich> Next job is to do the reverse, so the fiel atom gets written back...
[17:54] <cone-200> ffmpeg.git 03Paul B Mahol 07295218f53152: idcin: check chunk_size value before using it * 03http://tinyurl.com/cesesba03
[18:04] <cone-200> ffmpeg.git 03Paul B Mahol 07d7d5b5dfc1b6: flvdec: check return value of create_stream() * 03http://tinyurl.com/cn346do03
[18:04] Action: Daemon404 tries to figure out whats making fate ioc barf
[18:04] <Daemon404> also added a static uclibc build
[18:22] <TimNich> ffmpeg -codecs lists encoders: prores prores_anatoliy prores_kostya
[18:23] <durandal_1707> your point?
[18:29] <Tjoppen> oh yeah, those encoders don't share tables that they could share
[18:29] <Tjoppen> last time I looked
[18:29] <TimNich> because I'm trying to pin down a problem with it...
[18:35] <Daemon404> woo sigills
[18:42] <Daemon404> http://chromashift.org/log.txt
[18:42] <Daemon404> how is the sar denom getting set to 0?
[18:43] Action: Daemon404 also stares at args
[18:43] <Daemon404> saste_, this looks like your domain (filter graphs)
[18:46] <ubitux> michaelni: any reason to prefer the uint8_t casts over a simple & 0xff?
[18:46] <ubitux> you're worried about some optim or something?
[18:46] <ubitux> (avio patch)
[18:50] <Compn> ubitux : did you valgrind it ? 
[18:50] Action: Compn has no clue
[18:50] <ubitux> valgrind? oO
[18:50] <Compn> Tjoppen : arent they under different licenses?
[18:51] <ubitux> i'd better look at the assembly generated, but i suppose my gcc is smart enough for this kind of op
[18:51] <Compn> if so, how could they share things...
[18:51] <Compn> its verbotten! 
[18:52] <michaelni> ubitux, no reason its just bikeshed
[18:52] <ubitux> ok ok
[18:53] <michaelni> though one could probably argue uint8_t has a consistency benefit with int casts
[18:53] <michaelni> that is where signed is needed
[18:54] <cone-200> ffmpeg.git 03Tim Nicholson 070ee57f8b1476: rawdec.c: Extract interlace information from quicktime flag if it exisits. * 03http://tinyurl.com/bn55laa03
[18:54] <cone-200> ffmpeg.git 03Tomas Härdin 072b4bbd12c2ec: dvdec: Move the VSC pack parsing to before avctx->execute() * 03http://tinyurl.com/d55kuok03
[18:54] <cone-200> ffmpeg.git 03Tomas Härdin 077b383bd9d71c: dvdec: Set top_field_first from FS flag * 03http://tinyurl.com/crxqqqx03
[18:54] <cone-200> ffmpeg.git 03Tomas Härdin 078cbb8f53575d: dv: Set FS flag from AVFrame.top_field_first * 03http://tinyurl.com/cbjh62z03
[18:57] <durandal_1707> wtf is LVFF file? what create this?
[18:58] <Compn> no idea
[18:58] <Compn> people starting to make up samples for excuses to add code :P
[19:02] <Tjoppen> Compn: dunno
[19:03] <Tjoppen> you could have the one with a more restrictive license use code from the other perhaps? but not the other way around?
[19:05] <cone-200> ffmpeg.git 03Paul B Mahol 07bb502411ddb9: nutdec: check return value of av_new_packet() * 03http://tinyurl.com/d9gqw2j03
[19:05] <cone-200> ffmpeg.git 03Paul B Mahol 073ca8a2328878: lavf/audiointerleave: check return value of av_new_packet() * 03http://tinyurl.com/csbvtjj03
[20:02] <Daemon404> ok
[20:02] <Daemon404> 1b8f4d773c4d3bfcff9cdbc26200afbbaa445bc3 is what broke it
[20:02] <Daemon404> den is getting set to 0
[20:09] <michaelni> Daemon404, how can i reproduce it ?
[20:13] <Daemon404> michaelni, /configure --prefix=/home/daemon404/build_ffmpeg//install --samples=/home/daemon404/fate_ffmpeg/ --enable-gpl --enable-memory-poisoning --arch=x86_64 --cc=/home/daemon404/ioc/bin/clang --extra-cflags=-fcatch-undefined-c99-behavior --extra-ldflags='-fcatch-undefined-c99-behavior -lm' --disable-asm --enable-ftrapv --disable-optimizations
[20:13] <Daemon404> is what im doing
[20:13] <Daemon404> and tetsing with:
[20:13] <Daemon404> (normal clang should work i think)
[20:13] <Daemon404> fmpeg -nostats -cpuflags all -threads 1 -thread_type frame+slice -i /home/daemon404/fate_ffmpeg//cdxl/cat.cdxl -an -frames:v 16 -f framecrc -
[20:13] <Daemon404> ^ test
[20:23] <michaelni> Daemon404, Hmm and how is that related to "den is getting set to 0" ?
[20:23] <Daemon404> i traced the crash
[20:23] <Daemon404> [12:42] <@Daemon404> http://chromashift.org/log.txt
[20:23] <Daemon404> that commit causes that
[20:27] <michaelni> ok, ill look into it
[20:39] <cone-200> ffmpeg.git 03Nicolas Noirbent 07af32a1f2d1b8: doc/filters: pad examples should be a @subsection * 03http://tinyurl.com/cb9hvne03
[20:39] <cone-200> ffmpeg.git 03Michael Niedermayer 07547ec4d3034d: ffmpeg: check timebase validity before using it. * 03http://tinyurl.com/csedwnr03
[21:06] <Daemon404> michaelni, fixed
[21:06] <Daemon404> ioc is now happy again
[21:16] <Daemon404> added http://fate.ffmpeg.org/report.cgi?time=20121015191457&slot=x86_64-poky-linux-uclibc-static
[21:18] <ubitux> poky linux, kawaii~
[21:18] <Daemon404> i use to work on yocto/poky ;)
[21:20] <saste_> michaelni, do you have any idea for ticket #ticket 1801?
[21:20] <saste_> in other words, what's the correct way to set DTS
[21:21] <saste_> also, does it really belong to user-space?
[21:27] <cone-200> ffmpeg.git 03rogerdpack 07d9d547063648: docu: fix flite example * 03http://tinyurl.com/9p9jhpd03
[21:27] <cone-200> ffmpeg.git 03rogerdpack 077f5a78a0da8e: docu: add flite ffmpeg example * 03http://tinyurl.com/8mez4ns03
[21:27] <cone-200> ffmpeg.git 03rogerdpack 07a8eaa9ebc670: docu: change verb flite * 03http://tinyurl.com/97enst703
[21:27] <cone-200> ffmpeg.git 03rogerdpack 07dfd085aae2a1: docu: add word resize to scale filter, to make searching for the word resize simpler * 03http://tinyurl.com/9meslh603
[21:27] <cone-200> ffmpeg.git 03Michael Niedermayer 070cc3cd5c654d: Merge branch 'docu' of https://github.com/rdp/FFmpeg * 03http://tinyurl.com/99jf48l03
[21:30] <bryno> Are there any plans to expose the protocol API again?
[21:31] <saste_> bryno, that's a popular feature request, please file a ticket in trac (and let people upvote it)
[21:31] <bryno> I know that it was public at one point.  I made a lot of use of it ^_^
[21:34] <saste_> bryno, evil use of it? ;-)
[21:36] <michaelni> about tickt1801 i miss information to reproduce the problem, the timestamps should be set correctly in the correct timebase ...
[21:36] <cone-200> ffmpeg.git 03Stefano Sabatini 07a726ac9a11b4: examples/muxing: extend usage notice * 03http://tinyurl.com/cwrd2cj03
[21:40] <saste_> michaelni, cd doc/examples; make muxing; muxing out.mov
[21:41] <saste_> also what's the real meaning of "Encoder did not produce proper pts, making some up."?
[21:42] <ubitux> mp=geq is crashing :'(
[21:45] <j-b> morning
[21:49] <michaelni> saste_, that way one can only build muxing.c if one has updated all system libs to git master
[21:49] <michaelni> and headers
[21:49] <saste_> michaelni, no you can also use the source libraries
[21:49] <saste_> the compiled source dir libraries 
[21:50] <saste_> i still have to commit the README file
[21:50] <michaelni> ill look at this once its documented how to build muxing.c
[21:52] <j-b> michaelni: did we do what you wanted/needed?
[21:54] <michaelni> j-b, no iam still waiting for cone showing more complete commit messages, daily free pizza and world peace
[21:54] <j-b> more complete messsages? like the first line?
[21:54] <ubitux> and no more tinyurl :(
[21:55] <j-b> ok
[21:55] <j-b> will see what we can do
[21:58] <cone-200> ffmpeg.git 03Stefano Sabatini 0722c5cc239c25: tools/ffeval: include compat/getopt.c in case of missing system getopt() * 03http://tinyurl.com/9x2cy9k03
[22:00] <saste_> roger d pack = ?
[22:00] <saste_> is he on irc?
[22:03] <j-b> usually, he is not
[22:06] <saste_> should we use av_log or stderr in tools?
[22:45] <cone-200> ffmpeg.git 03Stefano Sabatini 07faa1cb50ed7d: lavfi/ass: extend syntax for ass filter * 03http://tinyurl.com/cbsx5ja03
[22:56] <Mista_D> anyway to debug a passlog file name that ffmpeg is looking for when encoding to two different files? "ratecontrol_init: can't open stats file" is all I get.
[00:00] --- Tue Oct 16 2012


More information about the Ffmpeg-devel-irc mailing list