[Ffmpeg-devel-irc] ffmpeg-devel.log.20181127

[burek]

[00:15:03 CET] <cone-786> ffmpeg 03Kyle Swanson 07master:fd2d6f376d84: libavf/libvmaf: update docs
[01:16:49 CET] <cone-786> ffmpeg 03Carl Eugen Hoyos 07master:5a7617136ab6: lavc/jrevdct: Avoid an aliasing violation.
[02:57:43 CET] <cone-786> ffmpeg 03Michael Niedermayer 07master:a036c2596949: avcodec/tiff: Fix integer overflows in left shift in init_image()
[03:47:22 CET] <cone-786> ffmpeg 03Jun Zhao 07master:e002ec332b75: lavfi/buffersrc: Indent the code.
[03:47:23 CET] <cone-786> ffmpeg 03Jun Zhao 07master:7608809ae424: fftools/ffprobe: Indent the code.
[03:47:24 CET] <cone-786> ffmpeg 03Jun Zhao 07master:1ffac23885ee: fftools/ffmpeg: delete the unused code.
[06:57:51 CET] <cone-015> ffmpeg 03Andrey Semashev 07master:1035206102e2: lavf/dashdec: Add webm to the list of allowed extensions.
[12:10:02 CET] <BtbN> What's it with all people going crazy over nvenc/cuda stuff lately? Just got a very upset E-Mail asking me to undo the nvenc unbundling, because it fucked streamers.
[12:10:58 CET] <durandal_1707> ?
[12:11:06 CET] <nevcairiel> the header unbundling?
[12:11:09 CET] <BtbN> yes
[12:11:21 CET] <nevcairiel> if they cant figure out how to get a second repository, they shouldnt be building code :p
[12:11:29 CET] <BtbN> It caused some distros to stop offering nvenc enabled ffmpeg builds, because it was auto-enabled before
[12:12:49 CET] <BtbN> But really, go complain to those distros, or just build it yourself.
[12:13:03 CET] <BtbN> All the more fancy stuff is non-free anyway
[12:18:46 CET] <thardin> I talked to a local streaming shop recently that does a lot of cuda stuff
[12:19:04 CET] <thardin> that and ~AI~ may be the reason
[12:28:53 CET] <nevcairiel> the yadif_cuda filter doesn't really need the cuda SDK, does it? Not at runtime? Only for compiling the kernel at build time?
[12:29:23 CET] <nevcairiel> hm i guess not all functions are in the compat header either
[13:02:26 CET] <durandal_1707> atomnuker: there is opus patch on ml that needs resolving
[13:10:51 CET] <BtbN> nevcairiel, the only thing making it non-free is it needing nvcc to build, yeah
[13:11:13 CET] <nevcairiel> but that doesnt really influence the binaries non-free status, does it
[13:11:20 CET] <BtbN> nope
[13:11:31 CET] <BtbN> (L)GPL demands that you are able to reproduce the build
[13:11:42 CET] <BtbN> which you clearly aren't without accepting Nvidias EULA
[13:11:49 CET] <nevcairiel> would be nice if a portable build without linked cuda sdk would be possible with th ose filters
[13:12:01 CET] <BtbN> It is
[13:12:05 CET] <BtbN> It's really only nvcc
[13:12:12 CET] <BtbN> to compile the .cu to .ptx
[13:12:25 CET] <nevcairiel> its not currently the way its setup, not without changes anyway
[13:12:44 CET] <BtbN> philipl added functions to the headers specifically for that though
[14:24:07 CET] <durandal_1707> atomnuker: what is status with opus decoding and gapless?
[14:39:08 CET] <cone-882> ffmpeg 03Michael Niedermayer 07master:28c96c2ce278: avcodec/diracdec: Check component quant
[15:24:40 CET] <durandal_1707> this is outrageous, all development power is sucked into dav1d
[15:25:35 CET] <j-b> durandal_1707: AV1 is the new kid in town. In a couple of months, it won't.
[15:25:48 CET] <j-b> and seeing the mess that AV1 is...
[15:45:35 CET] <kierank> durandal_1707: nonsense
[16:37:36 CET] <atomnuker> durandal_1707: should work, if the packet duration and length are correct
[18:38:37 CET] <cone-882> ffmpeg 03Carl Eugen Hoyos 07master:1893c7208602: lavfi/fps: Avoid duplicating Closed Captions when increasing frame rate.
[18:38:38 CET] <cone-882> ffmpeg 03Carl Eugen Hoyos 07master:698574ac5d68: ffmpeg: Avoid duplicating Closed Captions when increasing frame rate.
[18:49:29 CET] <cone-882> ffmpeg 03Carl Eugen Hoyos 07master:8cc5107705e9: lavf/Makefile: Fix demuxer dependencies: dvbsub, dvbtxt
[20:45:49 CET] <j-b> 'morning
[20:53:00 CET] <iive> where in the world is morning?
[20:53:18 CET] <iive> australia, japan?
[20:56:07 CET] <gnafu> iive: It is always morning when someone arrives on IRC, no?
[20:57:14 CET] <iive> that's strange notion.
[20:57:55 CET] <iive> I would prefer to think that j-b is enjoying the summer in Australia atm.
[21:40:54 CET] <durandal_1707> atomnuker: seen my opus patch?
[21:41:25 CET] <atomnuker> yeah, I'm not sure what it does
[21:42:36 CET] <atomnuker> I was sure gapless worked
[21:43:26 CET] <atomnuker> I guess if the packet doesn't have skip samples data decode.c uses avctx->internal->skip_samples
[21:43:36 CET] <atomnuker> so it should be okay
[21:44:40 CET] <atomnuker> but what is avctx->internal->skip_samples_multiplier
[21:45:42 CET] <atomnuker> apparently its only used if there's side packet data, but if there is what's the point of multiplying the value when you can directly set it and modify it
[21:48:18 CET] <atomnuker> ugh, its an aac he hack
[21:48:47 CET] <JEEB> sounded like it since IIRC it uses half the sample rate and the decoder then makes u for it or so?
[21:49:06 CET] <JEEB> and even in some containers the half-rates ended up in
[21:49:25 CET] <atomnuker> durandal_1707: anyway, looks fine, can you make a fate test for gapless opus? its missing from tests/fate/gapless
[22:36:15 CET] <j-b> iive: Australia, or NZ?
[22:37:36 CET] <iive> j-b, yeh, New Zealand is next to Australia.
[22:37:57 CET] <j-b> I would prefer imagining NZ then
[22:49:16 CET] <iive> deal. I'll imagine that one too.
[23:16:42 CET] <j-b> cool
[00:00:00 CET] --- Wed Nov 28 2018