[Ffmpeg-devel-irc] ffmpeg-devel.log.20171221

burek burek021 at gmail.com
Fri Dec 22 03:05:03 EET 2017


[04:15:57 CET] <cone-792> ffmpeg 03Steven Liu 07master:8318f60845bb: avformat/hlsenc: fix first fragment mp4 do not split bug
[04:19:55 CET] <cone-792> ffmpeg 03Steven Liu 07master:7feae7be5b08: avformat/hlsenc: reindent after previous commits
[04:40:56 CET] <cone-792> ffmpeg 03Karthick J 07master:3ac76d890d91: avformat/hlsenc: Fix a memory leak when http_persistent is 1
[04:50:50 CET] <cone-792> ffmpeg 03James Almer 07master:22ae33fb4083: configure: fix minimum required version of libzimg
[04:53:56 CET] <wm4> so what are the plans for the next ffmpeg release (based on current git master or later)
[04:57:51 CET] <jamrial> tagging it around late january or february, i guess
[04:58:23 CET] <jamrial> ffserver will be gone, and ideally so will the lavu atomic wrappers
[04:59:44 CET] <jamrial> and judging by the vote about winxp support, that will be gone too
[05:01:29 CET] <jamrial> maybe also the first version to have initial support for avx512 if jdarnley pushes his patchset
[05:03:46 CET] <wm4> I'll probably make a patch some time that introduces static mutex initializers to the windows pthread wrapper (which will come with a patch to remove the XP code)
[05:04:17 CET] <jamrial> that'd be a pretty good addition
[05:04:35 CET] <wm4> and then I want to deprecate the "lock manager"
[05:25:11 CET] <cone-792> ffmpeg 03James Almer 07master:bc10260f49d5: avformat/mov: simplify parsing of mdcv atom using av_make_q()
[11:21:56 CET] <atomnuker> wm4: oh yes please, that lockmgr code isn't good
[11:25:39 CET] <stevenliu> Hi guys
[11:25:43 CET] <stevenliu> The EXT-X-TARGETDURATION tag specifies the maximum Media Segment
[11:25:44 CET] <stevenliu>   duration.  The EXTINF duration of each Media Segment in the Playlist
[11:25:44 CET] <stevenliu>   file, when rounded to the nearest integer, MUST be less than or equal
[11:25:45 CET] <stevenliu>   to the target duration; longer segments can trigger playback stalls
[11:25:47 CET] <stevenliu>   or other errors.  It applies to the entire Playlist file.
[11:25:52 CET] <stevenliu>  how should i understand it?
[11:26:27 CET] <stevenliu> EXT-X-TARGETDURATION >= EXTINF? or EXT-X-TARGETDURATION can < EXTINF?
[11:29:50 CET] <stevenliu> i have sent a mail to http-live-streaming-review at group.apple.com for ask this question. i just curiosity the spec said EXT-X-TARGETDURATION must > EXTINF, but use mediastreamvalidate, it get error when EXT-X-TARGETDURATION > EXTINF, just when EXTINF is less than x.4444440, but if EXT-X-TARGETDURATION equ 4, it less than EXTINF.
[12:32:19 CET] <cone-121> ffmpeg 03Paul B Mahol 07master:c99ed89f8961: avfilter/af_biquads: add kHz width_type
[12:32:20 CET] <cone-121> ffmpeg 03Paul B Mahol 07master:7fc89f226fb6: avfilter/af_biquads: increase width range
[12:32:21 CET] <cone-121> ffmpeg 03Paul B Mahol 07master:5a6e753bc62f: avfilter/af_biquads: change defaults for biquad filter
[18:51:31 CET] <cone-395> ffmpeg 03wm4 07master:55eebf2a11d2: v4l_m2m: add missing AV_CODEC_CAP_DELAY flags
[18:51:31 CET] <cone-395> ffmpeg 03wm4 07master:1083859cb8c9: rkmppdec: move AV_CODEC_CAP_AVOID_PROBING to the correct field
[18:52:28 CET] <wm4> whoops, posted on the ML that I pushed these patches, but my stupid mail client only replied to the v4l2 guy
[19:13:30 CET] <nevcairiel> that usually happens to me if people answer to me directly and only CC the ML, which is kinda bad practice to begin with
[19:15:51 CET] <wm4> also I misspelled the prefix
[19:15:56 CET] <wm4> it's v4l2m2m or something
[19:16:06 CET] <wm4> (how did they manage to get two "2m" in there)
[19:16:47 CET] <wm4> the subsystem is called v4l2 because it was the second version of v4l, but the original v4l is long dead
[19:17:32 CET] <wm4> nevcairiel: btw. what's the best way to detect the presence of a Windows Vista+ target in configure?
[19:19:07 CET] <iive> m2m is memory to memory
[19:50:00 CET] <nevcairiel> wm4: check _WIN32_WINNT i guess
[19:50:14 CET] <nevcairiel> >= 0x0600 would be vista+
[19:50:45 CET] <wm4> hmmm I guess
[20:25:42 CET] <cone-395> ffmpeg 03Michael Niedermayer 07master:42274db1c623: avcodec/jpeg2000dec: Allocate lengthinc and data_start arrays as needed
[20:25:43 CET] <cone-395> ffmpeg 03Michael Niedermayer 07master:4b2a186ef02c: avcodec/mpeg4videodec: Add support for parsing and exporting video_range
[20:25:44 CET] <cone-395> ffmpeg 03Michael Niedermayer 07master:80344959f064: avcodec/jpeg2000dec: Free lengthinc earlier
[20:50:38 CET] <wm4> just ran configure on windows - took 24 minutes
[20:52:15 CET] <wm4> and I did it wrong, so I need to run it again
[20:55:27 CET] <tmm1> SortaCore: did you ever figure out a fix for those random qsv -17 errors?
[22:21:11 CET] <wm4> we're still in the ABI unstable phase right?
[22:21:29 CET] <jamrial> yeah
[22:21:52 CET] <nevcairiel> we should really get that sorted tho
[22:22:41 CET] <jamrial> the lavu atomic wrappers are still there
[22:22:56 CET] <jamrial> if atomnuker or someone else can't remove them, then we'll have to leave them there until the next bump
[22:23:50 CET] <wm4> changing this lock manager stuff allows us to remove an avpriv
[22:24:03 CET] <wm4> libavcodec defines this: avpriv_lock_avformat()
[22:24:10 CET] <wm4> and the only reason is this lock manager nonsense
[22:24:18 CET] <wm4> the WTFs are just piling up
[22:32:22 CET] <wbs> did you get usable static mutexes now? if you do, you can replace that with a private static mutex in libavformat
[22:32:49 CET] <wm4> wbs: we'll drop XP support
[22:33:08 CET] <wm4> Windows Vista provides a new kind of light weight lock (SRWs) which can be explicitly static
[22:33:15 CET] <wbs> yup
[22:33:25 CET] <wbs> does OS/2 have anything matching? :P
[22:33:29 CET] <wm4> who knows
[22:33:45 CET] <wm4> not my problem, and I think the OS/2 maintainer will take care of it
[22:33:47 CET] <wm4> I can CC him
[22:41:33 CET] <atomnuker> jamrial: Chloe was working on it and he was quite close
[22:42:09 CET] <Chloe> jamrial: when is next bump
[22:42:32 CET] <jamrial> who knows
[22:42:44 CET] <jamrial> could be in a year, in half a year, or in two years
[22:43:04 CET] <wm4> jesus christ ffplay uses the lock manager
[22:44:45 CET] <Chloe> wm4: ffmpeg on mingw (on Windows) is actually not practical at all
[22:46:04 CET] <atomnuker> Chloe: 2 years, so very long
[22:48:57 CET] <Chloe> I can just do it now I guess. 
[22:50:36 CET] <wm4> Chloe: well I used to do windows development in wine, but they don't support most newer APIs
[22:51:45 CET] <wbs> they do add more support for newer APIs all the time though, so it's worth trying with a very recent wine version
[22:52:04 CET] <wm4> I doubt I can hope for d3d11va and stuff
[22:58:50 CET] <wbs> I wouldn't count on that no
[23:08:28 CET] <cone-395> ffmpeg 03Marton Balint 07master:d6a8e46f9788: fate: add 12 bit framerate filter tests
[23:08:29 CET] <cone-395> ffmpeg 03Marton Balint 07master:1eb926dc0268: avfilter/vf_framerate: add threaded blending operations
[23:08:30 CET] <cone-395> ffmpeg 03Marton Balint 07master:090b740680f9: avfilter/vf_framerate: factorize blend_frames
[23:08:31 CET] <cone-395> ffmpeg 03Marton Balint 07master:be15304ee610: avfilter/vf_framerate: factorize get_scene_score
[23:08:32 CET] <cone-395> ffmpeg 03Marton Balint 07master:e1113a83ccb6: avfilter/vf_framerate: fix scene change detection score
[23:08:33 CET] <cone-395> ffmpeg 03Marton Balint 07master:c6a65ed67088: avfilter/vf_framerate: do not calculate scene change score multiple times for the same frame
[23:08:34 CET] <cone-395> ffmpeg 03Marton Balint 07master:e403e4bdbea0: avfilter/vf_framerate: fix scene score with negative linesize
[23:13:23 CET] <wm4> nice, windows restarted while compiling ffmpeg
[23:13:34 CET] <wm4> it's doing that for an update
[23:13:49 CET] <wm4> moronic OS
[23:14:35 CET] <BtbN> It usually bugs you about a needed restart for at least a day until it actually just does it
[23:18:45 CET] <nevcairiel> it does, and you can configure a time period you want to disable that stuff entirely,  usually that only happens when people ignore it for way too long
[23:19:40 CET] <BtbN> What sometimes happens is, that it does not perform the updates when you tell it to shutdown
[23:19:47 CET] <BtbN> But goes into hibernation instead
[23:19:58 CET] <BtbN> so after a few days of that, it does the actual reboot for the update
[23:20:24 CET] <BtbN> I have seen Windows with several days of uptime, even though they are "shut down" every night
[23:23:03 CET] <wm4> lol
[23:27:06 CET] <Chloe> has anyone tried the linux on windows thing for ffmpeg dev? how well does it work compared to a VM?
[23:29:05 CET] <wm4> that would still be cross compiling, so I imagine it's basically like a VM, with better filesystem sharing support
[23:29:48 CET] <nevcairiel> its a bit more odd then cross compiling, since you can basically run native win32 compilers in it and run the resulting binary as well
[23:30:22 CET] <wm4> that sounds like it actually has an advantage
[23:30:31 CET] <wm4> at least due to it using kernel fork
[23:30:48 CET] <nevcairiel> i havent really tried it much tho
[23:33:27 CET] <jdarnley> At least configure should be a bit quicker in it :D
[00:00:00 CET] --- Fri Dec 22 2017


More information about the Ffmpeg-devel-irc mailing list