[Ffmpeg-devel-irc] ffmpeg-devel.log.20190614

[burek]

[04:31:28 CEST] <cone-025> ffmpeg 03Zhong Li 07master:4208b428c473: lavf/qsv_vpp: add frame format option
[04:56:59 CEST] <cone-025> ffmpeg 03Steven Liu 07master:fa7a6dbd7607: avformat/hlsenc: add EXT-X-I-FRAMES-ONLY tag support
[04:57:01 CEST] <cone-025> ffmpeg 03Steven Liu 07master:aea524e6e325: avformat/dashdec: add http_proxy, referer and rw_timeout http method support for segments
[04:57:02 CEST] <cone-025> ffmpeg 03Steven Liu 07master:5d0652916f2f: avformat/hls: avformat_find_stream_info when the audio list in the variant
[04:57:02 CEST] <cone-025> ffmpeg 03Steven Liu 07master:bee0fba7e69f: avformat/hls: add http 2.0 version number compare for the http_multiple
[07:26:59 CEST] <cone-025> ffmpeg 03Jun Zhao 07master:c351873e06be: lavf/webvtt: fix ffmpeg -h demuxer=webvtt can't dump options
[07:27:00 CEST] <cone-025> ffmpeg 03Jun Zhao 07master:6f34fca8a8ad: doc/formats: document f_strict/strict for AVFormatContext
[07:27:01 CEST] <cone-025> ffmpeg 03Jun Zhao 07master:c432a91d252f: doc/fftools-common-opts: document ffmpeg -h bsf=bitstream_filter_name
[07:27:02 CEST] <cone-025> ffmpeg 03Jun Zhao 07master:6f82bf9dd30d: lavf/hls: More log message
[11:07:11 CEST] <cone-025> ffmpeg 03Steven Liu 07master:665759c3c803: avformat/dashdec: check copy_init_section memory alloc result
[12:41:35 CEST] <jidckii> Hello to all. Tell me, are there any problems at work https://trac.ffmpeg.org/wiki ? For several weeks, it opens very slowly.
[12:56:36 CEST] <jidckii> example https://i.imgur.com/mEBqwHD.png
[13:21:08 CEST] <durandal_1707> michaelni: there are huge issues with trac server, please fix it
[13:46:07 CEST] <cone-025> ffmpeg 03Limin Wang 07master:24321b8751b7: libavfilter/vf_blend.c: remove duplicate code with same function
[13:46:08 CEST] <cone-025> ffmpeg 03Limin Wang 07master:c9c1711f4961: libavfilter/vf_blend.c: remove duplicate code with macro-defined function
[18:10:20 CEST] <kierank> durandal_1707: good live resampler is hard
[19:33:08 CEST] <durandal_1707> kierank: fake!
[19:33:48 CEST] <kierank> durandal_1707: fix my lipsync problem, kthx
[19:34:22 CEST] <durandal_1707> kierank: use pcm
[19:34:45 CEST] <kierank> That's what I am doing
[19:34:49 CEST] <kierank> That is lipsync with pcm
[19:35:01 CEST] <kierank> Good resampling is hard
[19:35:07 CEST] <durandal_1707> than you are bad engineer
[19:35:21 CEST] <durandal_1707> resampling is trivial
[19:37:38 CEST] <kierank> Not if resample ratio is dynamic
[19:37:42 CEST] <kierank> And fractional
[19:49:50 CEST] <durandal_1707> kierank: you wrong, me correct
[19:53:24 CEST] <durandal_1707> Lynne: how is tx asm going?
[19:57:46 CEST] <durandal_1707> michaelni: why is release still not made?
[20:06:29 CEST] <michaelni> because the day has only 24h and theres always more work. For example trac today ...
[20:13:01 CEST] <JEEB> michaelni: btw since you were the only one who commented on https://patchwork.ffmpeg.org/patch/12492/ , it seems like the only thing wget/curl do is they have a "how much time until I/O timeout we have" in the loop. f.ex. https://github.com/curl/curl/blob/master/lib/vtls/gtls.c#L332
[20:13:13 CEST] <JEEB> I'm not 100% sure how we handle timeouts on this level, or if we do at all :P
[20:13:39 CEST] <JEEB> but it clearly seems like in normal usage (not thinking of I/O timeouts) we should just keep on looping on non-fatal errors
[20:14:23 CEST] <nevcairiel> thats definitely the documented API usage pattern, and timing out in the middle of a SSL handshake seems rather overkill
[20:14:25 CEST] <nevcairiel> so just go with i t
[20:15:02 CEST] <JEEB> yes, I looked at the API usage and docs and that's what the thing seems to be
[20:15:27 CEST] <JEEB> I think TLS handshakes have a limited amount of downgrades you can do etc so I would /guess/ in any sane case it will stop at some point or end up with a fatal error
[20:15:35 CEST] <JEEB> or well, that the list of ciphers is limited
[20:15:36 CEST] <JEEB> :P
[20:15:45 CEST] <JEEB> from client's and server's POV
[20:16:00 CEST] <JEEB> since that's what effectively you're looping over
[20:16:46 CEST] <nevcairiel> ship it
[20:19:35 CEST] Action: JEEB rebases locally
[20:21:24 CEST] <durandal_1707> michaelni: if you can't spend so much time with FFmpeg please say so!
[20:21:43 CEST] <kierank> durandal_1707: you should become leader
[20:21:49 CEST] <JEEB> durandal_1707: what's with your attitude recently?
[20:24:03 CEST] <durandal_1707> kierank: how?
[20:24:11 CEST] <kurosu> maybe some news he has been waiting for so long but isn't coming - growing restless
[20:24:36 CEST] <kierank> durandal_1707: just announce it
[20:24:42 CEST] <durandal_1707> lol
[20:29:23 CEST] <roxlu> hi, is this where libav dev. people hang out nowadays? 
[20:29:43 CEST] <JEEB> yes, the libraries' developers
[20:30:20 CEST] <roxlu> Ah cool ... maybe this question is a bit offtopic then (not sure)
[20:30:27 CEST] <roxlu> but I've experimented with using libav to decode video in the past and wanted to have a look at video + audio decoding. 
[20:30:54 CEST] <JEEB> there's some examples under doc/examples, but officially #ffmpeg is the "users' channel"
[20:30:58 CEST] <JEEB> ffmpeg.c or the APIs
[20:31:07 CEST] <roxlu> I'm curious how the API/design of libav fit's into the OS level audio callback. I mean what's the general approach?
[20:31:38 CEST] <roxlu> ok thanks JEEB. Should I ask this in #ffmpeg? 
[20:31:44 CEST] <JEEB> API usage, yup
[20:31:58 CEST] <roxlu> ok, thanks
[20:34:18 CEST] <cone-025> ffmpeg 03Remita Amine 07master:bc1749c6e460: lavf/tls_gnutls: retry gnutls_handshake on non fatal errors
[20:34:22 CEST] <JEEB> nevcairiel: there we go
[21:05:05 CEST] <nevcairiel> :)
[21:40:20 CEST] <cone-025> ffmpeg 03Michael Niedermayer 07master:6f2625aafc3d: tools/target_dec_fuzzer: Also fuzz extradata
[21:40:21 CEST] <cone-025> ffmpeg 03Michael Niedermayer 07master:8c6c4129b4cc: avformat/vpk: Fix integer overflow in samples_per_block computation
[21:40:22 CEST] <cone-025> ffmpeg 03Michael Niedermayer 07master:aa003019ab9e: avformat/vpk: Check offset for validity
[21:40:23 CEST] <cone-025> ffmpeg 03Michael Niedermayer 07master:2a0f23b9d647: avformat/sbgdec: Fixes integer overflow in str_to_time() with hours
[21:40:24 CEST] <cone-025> ffmpeg 03Michael Niedermayer 07master:e78b0f83748f: avcodec/bitstream: Check for integer code truncation in build_table()
[21:40:25 CEST] <cone-025> ffmpeg 03Michael Niedermayer 07master:a7e3b271fc9a: avcodec/bitstream: Check for more conflicting codes in build_table()
[21:40:26 CEST] <cone-025> ffmpeg 03Michael Niedermayer 07master:dd357d76e5fa: avformat/wtvdec: Avoid (32bit signed) sectors
[22:15:20 CEST] <mkver> jkqxz: I am currently working on the patches for vp9_raw_reorder and noticed a bug: Think about a frame that will never be displayed and also not referenced at all (so refresh_frame_flags is 0). The current code doesn't handle this scenario very well. It is insanse (the frame is essentially just filler material), but I haven't found anything that says that it is against spec, so I hesitate to simply error out on encountering 
[22:15:20 CEST] <mkver> such frames (although that's my preferred solution).
[00:00:00 CEST] --- Sat Jun 15 2019