[Ffmpeg-devel-irc] ffmpeg-devel.log.20120625

[burek]

[00:38] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r166f386446 10ffmpeg/tests/fate/vcodec.mak: 
[00:38] <CIA-119> ffmpeg: fate: speedup dnxhd tests and reduce their memory requirements
[00:38] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:39] <durandal_1707> finally
[02:04] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rf15803e957 10ffmpeg/tests/fate/acodec.mak: 
[02:04] <CIA-119> ffmpeg: fate: Try to fix source path for fate-acodec-dca
[02:04] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:48] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r5fd3e6965e 10ffmpeg/ (libavformat/vocenc.c tests/ref/lavf/voc_s16): 
[03:48] <CIA-119> ffmpeg: vocenc: use new header from codec tag 4
[03:48] <CIA-119> ffmpeg: this matches sox and should fix ticket1119
[03:48] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:48] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r0abfb0a9d8 10ffmpeg/ (libavformat/vocenc.c tests/lavf-regression.sh): 
[03:48] <CIA-119> ffmpeg: vocenc: change default codec to 16bit
[03:48] <CIA-119> ffmpeg: Hardly anyone would want 8bit today, 16bit is a much more reasonable
[03:48] <CIA-119> ffmpeg: default.
[03:48] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:48] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rbacbbd2b03 10ffmpeg/ (4 files in 3 dirs): 
[03:48] <CIA-119> ffmpeg: vocenc: fix sample rate rounding direction
[03:48] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[13:09] <brocatz> hey guys, i just compiled ffmpeg with the mingw package so that i could apply a patch and everything works fine, but now when i try to use avisynth it fails, it says permission denied, but i think that error is a side effect of another error, my other version of ffmpeg from 2010 (when i last used avisynth with it) works fine
[13:10] <brocatz> wondering if there are any obvious things i should check
[13:10] <brocatz> er w/w
[13:24] <Compn> find out what permission is failing
[13:24] <Compn> probably script in wrong dir
[13:25] <brocatz> the exact command works on an older binary
[13:25] <brocatz> i have 3 binaries, a recent nightly, my build, and an older one
[13:25] <brocatz> swapping them out, so everything else is the same, the older one runs fine
[13:25] <brocatz> [avs @ 00000000016cf320] AVIFileOpen failed with error -2147221164
[13:25] <brocatz> C:/temp/output.mp4.1.avs: Operation not permitted
[13:25] <brocatz> that's the error btw
[13:25] <Compn> oh
[13:26] <Compn> well then something may have broken, or changed, since then
[13:26] <Compn> update your script to new avisynth syntax ? 
[13:26] <brocatz> i'm using the Version avisynth script
[13:26] <brocatz> which is very basic
[13:29] <brocatz> hum the newest nightly i just got is fine
[13:29] <brocatz> could i be inadvertently compiling my ffmpeg as 64bit
[13:32] <brocatz> http://dpaste.com/763205/ that's my configure string
[13:38] <brocatz> maybe someone could integrate our small patch so i can just use the main version of ffmpeg :)
[13:45] <Compn> what patch ?
[13:46] <brocatz> https://ffmpeg.org/trac/ffmpeg/ticket/1463
[13:48] <Compn> ah yest hat one
[13:49] <michaelni> brocatz, if i understood reimar correctly that would break support for pre win2k ?
[13:50] <michaelni> that is a binary compiled on win2k wouldnt work on previous versions ?
[13:51] <michaelni> while one compiled on previous versions would have non working ctrl-break on later
[13:51] <piman> The patch doesn't break support for anything.
[13:52] <piman> It adds support for Ctrl+Break on Windows >= 2k.
[13:52] <piman> (Hi, I'm the patch author.)
[13:52] <michaelni> hi piman 
[13:52] <piman> Is pre-2k Windows a legitimate target for you guys still?
[13:52] <michaelni> honestly i dont know :)
[13:53] <michaelni> but if it works before your patch then i would prefer it still does afterwards
[13:54] <michaelni> if it doesnt then the configure check shouldnt be needed
[13:54] <piman> I'd think any problems it introduces are identical to whatever problems are introduced by the use of e.g. GetProcessMemoryInfo which is done in the same way my patch is.
[13:55] <piman> (And has been there for years, and is XP+ only)
[13:57] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * r972cad77fa 10ffmpeg/libavfilter/ (11 files): (log message trimmed)
[13:57] <CIA-119> ffmpeg: lavfi: remove unnecessary inclusion of libavcodec/avcodec.h in avfilter.h
[13:57] <CIA-119> ffmpeg: libavfilter API was designed in order to be clarly distinguished from the
[13:57] <CIA-119> ffmpeg: libavcodec API, including avcodec.h in avfilter.h is not going to help to
[13:57] <CIA-119> ffmpeg: stick to this principle.
[13:57] <CIA-119> ffmpeg: The inclusion of libavutil/audioconvert.h in many files was required
[13:57] <CIA-119> ffmpeg: because avcodec.h includes audioconvert.h.
[13:57] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * rfec512a52c 10ffmpeg/ffplay.c: ffplay: give more meaningful names to the buffersink instances
[13:57] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * r43583fb85c 10ffmpeg/ (5 files in 2 dirs): (log message trimmed)
[13:57] <CIA-119> ffmpeg: lavfi/avcodec: deprecate avfilter_fill_frame_from_*_buffer_ref API
[13:57] <CIA-119> ffmpeg: Deprecate functions:
[13:57] <CIA-119> ffmpeg: avfilter_fill_frame_from_buffer_ref
[13:57] <CIA-119> ffmpeg: avfilter_fill_frame_from_audio_buffer_ref
[13:57] <CIA-119> ffmpeg: avfilter_fill_frame_from_video_buffer_ref
[13:57] <piman> If you'd rather the configure check be gone I can change it to #ifdef _WIN32 but I don't really see the benefit in that. I usually prefer to test for the features I need rather than platforms.
[13:57] <CIA-119> ffmpeg: and schedule to drop them at the next API major bump.
[13:57] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * ra31ab50712 10ffmpeg/libavfilter/ (avfilter.h version.h): 
[13:57] <CIA-119> ffmpeg: lavfi: move some FF_API_ definitions from avfilter.h to version.h
[13:57] <CIA-119> ffmpeg: version.h seems the right place for FF_API_ definitions.
[13:58] <piman> (e.g. maybe Windows 9 RT stops defining _WIN32 finally, my check would still work.)
[13:59] <michaelni> piman, are versions prior 2k still supported by MS ? i mean in terms of security updates ?
[14:00] <piman> The usual policy for that kind of stuff for a company that size is "if you pay us enough, we'll do it"
[14:00] <piman> But no, it's not generally supported at all anymore.
[14:01] <brocatz> XP Support is about to end too, April 2014
[14:01] <michaelni> i guess then the patch might be ok, let me take another look at it 
[14:02] <burek> too bad.. it's the best OS they've ever had
[14:02] <piman> What exactly is the concern? I guess you're thinking someone compiles on Win7, gets a binary, and it doesn't run on Windows 98?
[14:03] <michaelni> yes 
[14:03] <piman> But that's going to happen anyway because of GetProcessMemoryInfo, and the fact every Windows compiler generates i686 instructions, etc.
[14:04] <michaelni> i was unaware of the  GetProcessMemoryInfo issue, this maybe should be changed somehow if its possible
[14:04] <Tjoppen> I find it interesting old 3.1 programs still work in XP
[14:04] <Tjoppen> or clock.exe did at least
[14:05] <Compn> michaelni : no, win2k is not supported anymore
[14:05] <Compn> winxp is due to expire soon (next few years iirc)
[14:06] <Compn> oh brocatz mentioned this already :P
[14:06] <brocatz> most of the people i know doing ffmpeg + windows work use virtual servers or real servers running 2008 server
[14:06] <Compn> virtual really took off lately
[14:07] <Compn> multiple cores did it i think
[14:15] <michaelni> piman, #ifdef are wrong, these things get defined to 0 so compile fails with the patch
[14:24] <michaelni> piman, if you resubmit the patch with #ifdef fixed ill apply it in a day or 2 if noone objects
[14:25] <piman> OK. To Trac or the mailing list?
[14:25] <michaelni> want to give reimar a chance to comment
[14:25] <michaelni> ML is better
[14:28] <piman> OK, thanks.
[14:28] <michaelni> np
[14:46] <Nedwada> hello guys
[14:46] <Nedwada> 'm having a problem, using ffmpeg version 0.10.2.git.   If i open an m3u8, and this is redirected inside its stream, i get a segfault error
[14:47] <Nedwada> http://privatepaste.com/2374deb5f0
[14:53] <Nedwada> can someone help?
[16:39] <burek> just to say this and wont prolong the topic discussed there on #ffmpeg :) please take a look at http://doc-book.sourceforge.net/homepage/ and if it's a viable solution I can convert all the texi into XML files
[16:39] <burek> so we can resolve that issue once for good
[16:49] <burek> DocBookWiki can also be used to edit a DocBook document online, from the web. Editing is done one section at a time, so the editor selects first the section that he wants to edit, and then edits it. He can edit it in several modes: text (like wiki), xml (the original format), html, latex, texi, etc., whichever is more suitable for him. The changes, however, are always saved in the XML(DocBook) format. Authentication of the editors (with username, pas
[16:49] <burek> sword) can be enabled as well, if necessary. Also, different editing permissions can be assigned to editors.
[16:50] <burek> so, devels edit xml files in text editor (offline), submit it through git (just like they use with texinfo files)
[16:50] <burek> with the difference that those same xml files can be nicely edited from the web, by regular users (i.e. not developers)
[17:24] <ubitux> https://lists.libav.org/pipermail/libav-devel/2012-June/029761.html
[17:24] <ubitux> this might interest ffmpeg dev as well
[17:26] <saste> http://sophia.estec.esa.int/socis2012/
[17:27] <saste> which reminds me that we still didn't integrated the patch from socis 2011
[17:27] <saste> *patches
[17:48] <michaelni> ubitux, ffmpeg will keep r_frame_rate, its essential for supporting some containers
[17:49] <ubitux> can you explain a bit?
[17:49] <michaelni> avi
[17:49] <ubitux> i don't see anything directly related in avi muxer
[17:49] <ubitux> (or demuxer)
[17:50] <ubitux> or maybe it's calling one of the lavf/utils
[17:52] <michaelni> iam not sure what i should explain, avi needs r_frame_rate or equivalent
[17:53] <michaelni> if eleril removes it in libav thats just another feature we have more :)
[17:53] <av500> for avi average frame rate == frame rate
[17:53] <michaelni> no
[17:53] <av500> no?
[17:54] <michaelni> avi supports skiped frames
[17:54] <michaelni> so its vfr strictly speaking
[17:54] <michaelni> just not efficiently
[17:54] <av500> well, strictly speaking it does not I would say
[17:54] <av500> people put 0 sized frames inside it
[17:55] <av500> but anyway, no point to argue that
[17:57] <CIA-119> ffmpeg: 03Ronald S. Bultje 07master * r8123e0901f 10ffmpeg/ (11 files in 3 dirs): 
[17:57] <CIA-119> ffmpeg: x86: place some inline asm under #if HAVE_INLINE_ASM
[17:57] <CIA-119> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[17:57] <CIA-119> ffmpeg: 03Mans Rullgard 07master * r8299260470 10ffmpeg/libavcodec/x86/ (Makefile fft_mmx.asm fft_sse.c): x86: fft: convert sse inline asm to yasm
[17:57] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * ra6ff8514a9 10ffmpeg/: (log message trimmed)
[17:57] <CIA-119> ffmpeg: Merge remote-tracking branch 'qatar/master'
[17:57] <CIA-119> ffmpeg: * qatar/master:
[17:57] <CIA-119> ffmpeg:  wtv: Check the return value from gmtime
[17:57] <CIA-119> ffmpeg:  x86: fft: convert sse inline asm to yasm
[17:57] <CIA-119> ffmpeg:  x86: place some inline asm under #if HAVE_INLINE_ASM
[17:57] <CIA-119> ffmpeg: Conflicts:
[17:57] <CIA-119> ffmpeg: 03Martin Storsjö 07master * rdc53858063 10ffmpeg/libavformat/wtv.c: 
[17:57] <CIA-119> ffmpeg: wtv: Check the return value from gmtime
[17:57] <CIA-119> ffmpeg: On MSVC, gmtime returns NULL for values outside of their supported
[17:58] <michaelni> av500, btw i got the old beagle board half working :)
[17:58] <michaelni> needed a powered usb hub for the lan usb to work
[17:58] <michaelni> and the timing of connecting the hub has to be right 
[17:59] <michaelni> connect it too early and nothing will work on the otg usb
[17:59] <av500> ok
[17:59] <av500> yes, that otg is a pain
[17:59] <av500> fighting it here at work since years :(
[18:00] <j-b> beagle is the one in omap3?
[18:00] <av500> yes
[18:01] <michaelni> about r_frame_rate, if someone has an idea on how to do avi vfr without it iam very interrested
[18:01] <michaelni> OTOH if this is just removial of a features then iam not and it sounds like it is
[19:21] <durandal_1707> ubitux: when you gonna remove SVN from download page?
[19:21] <ubitux> gonna push tonight
[19:21] <ubitux> i want to do something about Compn comment first
[19:22] <ubitux> and i will likely update the documentation page as well (to add the wiki URL at least)
[19:23] <saste> durandal_1707: are you going to update your smtpebars source? or do you want me to refactor it?
[19:23] <durandal_1707> saste: not any time soon, i'm working on stupid undocumented video codec(s)
[19:23] <ubitux> saste: what is blocking you from submitting your segmenter branch? :)
[19:23] <ubitux> durandal_1707: smush ?
[19:24] <durandal_1707> ubitux: san variant
[19:24] <ubitux> ok
[19:24] <ubitux> durandal_1707: btw you forgot the @tab X @tab X
[19:24] <ubitux> in general.texi
[19:25] <ubitux> (put the X where you need to, but the @tab are missing, and output is "broken")
[19:25] <ubitux> (look for Smush on http://ffmpeg.org/general.html#File-Formats)
[19:25] <durandal_1707> omg
[19:38] <CIA-119> ffmpeg: 03Paul B Mahol 07master * rebfcd6049f 10ffmpeg/doc/general.texi: 
[19:38] <CIA-119> ffmpeg: doc/general: fix output
[19:38] <CIA-119> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[19:48] <ubitux> durandal_1707: thx :)
[21:16] <CIA-119> ffmpeg: 03Carl Eugen Hoyos 07master * rbec21ce7f4 10ffmpeg/libavcodec/mss1.c: Make MSS1 palette opaque.
[22:04] <durandal_1707> is there filter which merges multiple audio streams into single stream?
[22:16] <ubitux> amerge?
[22:23] <ubitux> yay, some free time at least.
[22:23] <ubitux> (last*?)
[22:29] <ubitux> Compn: i think the git clone should be before the snapshot url...
[22:49] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * r8a3544149f 10ffmpeg/libavfilter/avfilter.h: 
[22:49] <CIA-119> ffmpeg: lavfi/avfilter.h: reorganize headers disposition
[22:49] <CIA-119> ffmpeg: Group lavfi headers together, slightly improve readability.
[22:49] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * r6be8cfa034 10ffmpeg/libavfilter/buffersrc.c: 
[22:49] <CIA-119> ffmpeg: lavfi/abuffer: increase logging level of the log showing initial parameters
[22:49] <CIA-119> ffmpeg: Also show them in a more parsable/consistent fashion.
[22:49] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * r9b41ec4b9e 10ffmpeg/libavfilter/vf_drawtext.c: 
[22:49] <CIA-119> ffmpeg: lavfi/drawtext: use av_opt_free() to free private context
[22:49] <CIA-119> ffmpeg: Simplify code.
[22:49] <RobertNagy> shouldn't the atempo filter change the time_base of the fitler graph?
[22:50] <RobertNagy> as it is now time moves at the same rate as without atemp
[22:50] <RobertNagy> which causes video to run at it's normal rate when using audio as master clock
[22:54] <ubitux> saste: check if it passes make checkheaders
[22:55] <ubitux> (about your stddef.h include)
[22:56] <saste> ubitux: I moved it up, how can that break something?
[22:56] <ubitux> i was refering to the original patch :)
[23:00] <saste> RobertNagy: the *pts* is rescaled, not the time_base
[23:01] <RobertNagy> the pts I get out is wrong then
[23:01] <RobertNagy> and why scale the pts? isn't it more accurate to change the time_base?
[23:02] <RobertNagy> I have to do "av_q2d(time_base) * tempo" to get the correct time
[23:02] <RobertNagy> though maybe I'm the one that's wrong
[23:05] <saste> RobertNagy: time_base is conveniently expressed as 1/sample_rate, why to change that?
[23:06] <Daemon404> 1/sample_rate seems like the best idea...
[23:06] <RobertNagy> well the problem I have is that, even if i set temp=0.5, time moves at the same rate as before
[23:07] <RobertNagy> I would expect time to move 0.5 times slower
[23:07] <RobertNagy> so that when I sync the video clock with the audio clock
[23:07] <RobertNagy> the video will got at the same rate as the audio
[23:07] <RobertNagy> though I'm probably missunderstanding how the atempo filter is ment to be used
[23:08] <durandal_1707> hmm, audio and video are separate streams
[23:08] <RobertNagy> yes? I still need to sync them
[23:10] <durandal_1707> for video stream you would use another filter...
[23:11] <RobertNagy> now you lost me, what does this have to do with the video filters?
[23:15] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rcd6aa95caa 10ffmpeg/libavcodec/h264_loopfilter.c: 
[23:15] <CIA-119> ffmpeg: h264_loopfilter: use av_assert
[23:15] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:15] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rdc30c27eb1 10ffmpeg/libavcodec/h264_parser.c: 
[23:15] <CIA-119> ffmpeg: h264_parser: use av_assert
[23:15] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:18] <durandal_1707> amerge can merge only 2 streams into one
[23:18] <Daemon404> can you chain multiple amerges?
[23:18] <durandal_1707> lol, description is ffmpeg output is wrong
[23:18] <durandal_1707> doc/filters mentions n streams
[23:22] <saste> RobertNagy: how are you synching the two streams (A+V)?
[23:23] <RobertNagy> audio_time = audio_frame->pts * audio_filter->time_base
[23:23] <saste> or in other words, are you doing it programmatically or through ff* tools?
[23:23] <RobertNagy> programmatically
[23:24] <RobertNagy> video_time = video_frame->pts * video_filter_->time_base
[23:24] <RobertNagy> while(video_time < audio_time) next_frame();
[23:24] <RobertNagy> something like that
[23:26] <RobertNagy> while(video_time < audio_time) next_video_frame();
[00:00] --- Tue Jun 26 2012