[Ffmpeg-devel-irc] ffmpeg-devel.log.20170228

[burek]

[00:23:51 CET] <J_Darnley> jamrial: possibly but I was looking at the INIT_MMX and INIT_XMM stuck into the middle
[00:45:18 CET] <kierank> michaelni: any comments on my eac3dec review?
[00:48:23 CET] <kierank> ah ignore me
[01:15:48 CET] <cone-188> ffmpeg 03Michael Niedermayer 07master:4ea7744859dc: avcodec/h264idct_template: Fix multiple runtime error: signed integer overflow
[01:15:49 CET] <cone-188> ffmpeg 03Michael Niedermayer 07master:25e93aacc214: avcodec/mpeg4videodec: Fix runtime error: left shift of negative value -2650
[01:15:50 CET] <cone-188> ffmpeg 03Michael Niedermayer 07master:63e400a8807d: avcodec/pictordec: Check plane value before doing value/mask computations
[01:15:50 CET] <cone-188> ffmpeg 03Michael Niedermayer 07master:949d2176ef0a: avcodec/dca: Fix multiple runtime error: signed integer overflow
[01:15:51 CET] <cone-188> ffmpeg 03Michael Niedermayer 07master:4c0139463c8f: avcodec/vp56: Factorize vp56_render_mb() out
[01:15:53 CET] <cone-188> ffmpeg 03Michael Niedermayer 07master:3d66eb6ec333: MAINTAINERS: Make formating of ffmpeg-security entry the same as others
[01:15:54 CET] <cone-188> ffmpeg 03Michael Niedermayer 07master:4f218682071c: MAINTAINERS: add wm4 to ffmpeg-security
[01:15:55 CET] <cone-188> ffmpeg 03Michael Niedermayer 07master:7f62368bacdb: MAINTAINERS: Add Rodger Combs to ffmpeg-security
[02:00:35 CET] <jpabq> What is the current state of support for CEA-608/708 ?   I have a patch to add muxing of captions into transport stream, but it against an OLD version of ffmpeg.  It was written by an ex-coworker, and I am far from an expert on the code.  I am willing to work on getting it committed, if ffmpeg is still lacking that functionality.
[02:12:36 CET] <iive> how old?
[02:17:21 CET] <Compn> jpabq : for mpeg-ts ?
[02:19:24 CET] <Compn> jpabq : probably we dont have support for it in mpegts
[02:20:39 CET] <Compn> jpabq : this bug i think https://trac.ffmpeg.org/ticket/1778
[03:46:50 CET] <jamrial> kierank: you may be interested in my latest patch if you were looking for h264 optimizations using bmi2
[03:47:59 CET] <jamrial> not sure how much impact it will have. it depends on how much get_bits() is used
[04:17:24 CET] <atomnuker> I think it might be worth removing support for soxr
[04:18:33 CET] <atomnuker> did some benchmarks, it seems lswr has the same quality and is faster
[12:02:41 CET] <jkqxz> Did anyone else test libturing?
[12:18:51 CET] <wm4> not me
[13:31:04 CET] <pszemus> Hi, how long does it take for a libavformat patch to be reviewed? I'm waiting 6 days now and I wonder whether to do something else.
[13:31:26 CET] <JEEB> if no replies, ping
[13:35:47 CET] <pszemus> I've pinged the mailing list 1 day after submitting the patch (5 days ago).
[13:36:01 CET] <pszemus> Still no replies/
[13:36:02 CET] <JEEB> well yeah :P
[13:36:31 CET] <JEEB> what was your patch about?
[13:39:47 CET] <pszemus> http://ffmpeg.org/pipermail/ffmpeg-devel/2017-February/207461.html
[13:40:38 CET] <JEEB> I just wanted the general gist so I knew if I could look at it or not :P
[13:40:53 CET] <JEEB> as looking at code I can't really have a say on isn't really useful :D
[13:41:22 CET] <pszemus> It's about adding the ability to change the bitrate/bandwidth field inside DASH manifest file.
[13:45:23 CET] <JEEB> I remember changing that in movenc's ISML manifests recently
[13:45:31 CET] <pszemus> I was looking for a way to set the bitrate/bandwidth field inside DASH manifest file AFTER encoding that's not CBR.
[13:45:41 CET] <JEEB> if bit rate is set, use that, if it's not set use maxrate
[13:46:02 CET] <JEEB> http://git.videolan.org/?p=ffmpeg.git;a=commit;h=6c10f8fe7658c699fc41688c175aa9d79f4b1daf
[13:46:11 CET] <JEEB> and I thought I took that out of dashenc?
[13:47:50 CET] <wbs> JEEB: there's similar changes (plus lots more) for dashenc in libav, if someone cares to merge/cherrypick
[13:48:12 CET] <pszemus> I wanted it to be accurate, so I'm computing each stream's bitrate after the encoding (after writing last packet) and I want to set that value inside DASH manifest
[13:48:47 CET] <JEEB> right
[13:48:48 CET] <pszemus> that's what my patch is about
[13:49:10 CET] <JEEB> I will take a look at after work
[13:49:29 CET] <pszemus> setting a'priori max_bitrate or bitrate is not a solution for me
[13:49:33 CET] <pszemus> thanks
[13:49:46 CET] <pszemus> (be back in 20min)
[13:49:50 CET] <JEEB> wbs: dashenc isn't currently something I'm utilizing so I will move onto that after the reviews on libav's side for your 2nd version of negative cts offsets
[13:50:13 CET] <JEEB> wbs: just fyi the negative cts offsets minipatch fixup has been running on a test box since Friday
[13:50:19 CET] <JEEB> so I'd guess it seems good :)
[13:50:23 CET] <wbs> JEEB: \o/
[13:50:51 CET] <wbs> JEEB: I've got another idea to make things even better/simpler, but I want the day to have more hours to actually implement it :P
[13:51:21 CET] <JEEB> I'll re-test quickly with the v2 patches since they contain a little bit of extra changes, but expect another LGTM from me during today :)
[15:26:30 CET] <durandal_1707> kierank: how is interlaced signaled in cfhd?
[15:27:19 CET] <kierank> probably a tag somewhere
[15:27:27 CET] <kierank> you have to print all the tags and compare
[15:45:13 CET] <durandal_1707> kierank: there are to many unparsed tags, perhaps it is mentioned in specification?
[15:45:50 CET] <kierank> no it is not in spec
[15:45:52 CET] <kierank> you have to compare tags
[18:01:14 CET] <carrybubbles> Hello guys, I have a question about Qualification Task: XPM decoder implemented in libavcodec in project "FITS image decoder and encoder". Do i have to decode one picture or video(many pictures)?
[18:03:45 CET] <Compn> carrybubbles : checking
[18:04:12 CET] <durandal_1707> carrybubbles: xpm files are single pictures
[18:04:41 CET] <Compn> maybe hes also asking about if FITS is image or also video...
[18:05:14 CET] <carrybubbles> nice rofl bro 
[18:07:40 CET] <durandal_1707> FITS format have specification freely available
[18:10:14 CET] <sfan5> btw what about the openssl 1.1.0 patches (configure + rtmpdh)
[18:10:30 CET] <sfan5> i don't care whether you cherry-pick the second one from Libav
[18:10:35 CET] <sfan5> would just be nice to have it working
[18:21:21 CET] <RiCON> the one from libav will be merged eventually
[18:39:30 CET] <adeeln_> durandal_170: getting started on the probe thing
[18:40:49 CET] <durandal_170> adeeln_: add xpm_pipe demuxer with probing
[18:54:40 CET] <adeeln_> Should I follow something similar to this [https://ffmpeg.org/pipermail/ffmpeg-devel/2014-January/153614.html]?
[19:00:22 CET] <durandal_170> adeeln_: no, see existing image demuxers
[19:00:30 CET] <adeeln_> okay
[19:12:33 CET] <adeeln_> durandal_170: A bit lost.Going through this http://ffmpeg.org/pipermail/ffmpeg-devel/2016-June/195680.html where would I be adding this?
[19:14:29 CET] <durandal_170> adeeln_: just add entry to libavformat/img2.c
[19:14:38 CET] <durandal_170> for XPM
[19:15:47 CET] <durandal_170> to ff_img_tags
[19:16:54 CET] <adeeln_> it's already there
[19:23:07 CET] <adeeln_> durandal_170: What should I do? I don't think there's any need for re-compiling
[19:24:09 CET] <durandal_170> adeeln_: is your xpm file having xpm extension?
[19:25:31 CET] <adeeln_> yes, I named it 'test.xpm'
[19:27:42 CET] <durandal_170> adeeln_: are you on linux?
[19:28:06 CET] <adeeln_> Yes.
[19:29:29 CET] <durandal_170> adeeln_: post link to gist what you have
[19:30:13 CET] <adeeln_> the output of 'ffprobe -h decoder=xpm_pipe'?
[19:31:40 CET] <durandal_170> adeeln_: no, code diff
[19:33:32 CET] <adeeln_> durandal_170: there you go: https://gist.github.com/adl1995/5a454e565c5c053cc5908526d57b1f18
[19:37:01 CET] <durandal_170> adeeln_: more like output of: git diff master
[19:45:46 CET] <adeeln_> durandal_170: they both have same size. should I commit first?
[19:46:49 CET] <durandal_170> adeeln_: yes, commit. but in separate branch
[19:55:03 CET] <adeeln_> durandal_170: do I commit and push to separate branch? or is there a way to commit to another branch?
[19:56:31 CET] <J_Darnley> git checkout -b NAME will create and switch to a new branch
[19:58:29 CET] <J_Darnley> Then you can make your commits in that new branch
[20:00:54 CET] <atomnuker> why bother with branches at all?
[20:01:03 CET] <atomnuker> it all goes to master anyway
[20:03:38 CET] <adeeln_> atomnuker: durandal_170 told me to create a difference for the two, J_Darnley: thank you 
[20:32:38 CET] <J_Darnley> Should the fate-h264-api test be respecting to the CPUFLAGS make option?
[20:32:41 CET] <J_Darnley> Can anyone else report that it is *not* respecting it?
[20:33:15 CET] <kierank> might be the case
[20:35:18 CET] <J_Darnley> Oh, I guess it doesn't looking at its source.
[21:53:50 CET] <cone-306> ffmpeg 03Carl Eugen Hoyos 07master:b8a7dcbde2e1: lavf/matroskadec: Fix demuxing sipr.
[00:00:00 CET] --- Wed Mar  1 2017