[Ffmpeg-devel-irc] ffmpeg-devel.log.20150907

[burek]

[01:02:49 CEST] <BBB> microchip_: probably a bug, send a patch
[02:36:57 CEST] <cone-318> ffmpeg 03Michael Niedermayer 07master:b480f0e37a42: avformat/cafenc: Move special casing of PCM codecs into ff_codec_caf_tags
[04:31:57 CEST] <cone-318> ffmpeg 03tab 07master:fb135139fd39: avformat: implement query_codec for the image2 muxer.
[07:23:54 CEST] <rcombs> should the image2 muxer have the AVFMT_VARIABLE_FPS flag, so it gets VSYNC_PASSTHROUGH by default instead of VSYNC_CFR?
[09:19:57 CEST] <compn> wonder what dell uses rtmpdump for
[10:28:24 CEST] <nevcairiel> ubitux: re avopt bool, i assume if someone in code uses av_opt_set_int on  such a bool option today, it'll continue to work in the future?
[10:28:52 CEST] <ubitux> i think so, but i'll test
[10:38:21 CEST] <nevcairiel> (or av_opt_set with "0" as the string, for that matter)
[10:38:53 CEST] <nevcairiel> but that seems to be explicitly covered, i guess
[10:52:01 CEST] <ubitux> yeah the "0" and "1" might actually not be needed, but i'll check
[10:52:27 CEST] <ubitux> on/off is a good one i forgot
[10:52:40 CEST] <saste> michaelni nevcairiel ubitux: what about if we have an IRC meeting this saturday?
[10:52:51 CEST] <ubitux> sure
[10:53:08 CEST] <nevcairiel> fine with me
[10:53:26 CEST] <saste> since we're going to discuss important matters at VDD and some important devs won't be here I think it's good if we can discuss it online
[10:53:48 CEST] <saste> something like 16 UTC is fine with you?
[10:55:32 CEST] <nevcairiel> any EU friendly time works for me
[10:55:48 CEST] <ubitux> ok with me
[10:57:56 CEST] <ubitux> saste: https://framadate.org/infos_sondage.php (doodle like)
[10:58:56 CEST] <saste> ubitux, is that to settle the meeting time?
[10:59:08 CEST] <ubitux> it's a poll
[10:59:28 CEST] <ubitux> you can use it for that typically
[10:59:41 CEST] <ubitux> (might have better alternatives, just picked a random one)
[11:00:00 CEST] <ubitux> (multi choice poll)
[11:29:10 CEST] <saste> any idea for the topics of the day of the IRC meeting?
[11:29:53 CEST] <saste> i'm not very involved with ffmpeg development lately, so i'm not the best to propose one
[11:30:53 CEST] <ubitux> saste: ABI compat policy, general policy decision process, VDD15
[11:31:59 CEST] <saste> ubitux, thanks
[11:38:44 CEST] <saste> allright, I sent the email but feel free to suggest a better meeting time or topics to discuss
[11:57:32 CEST] <ubitux> from a technical PoV, i think we might want to discuss HEVC asm (poke BBB), the usual lavfi limitations, what are the remaining cleanup priorities 
[12:25:58 CEST] <cone-320> ffmpeg 03Federico Tomassetti 07master:6bcd3e05998b: build: Add support for travis CI
[12:25:59 CEST] <cone-320> ffmpeg 03Vittorio Giovara 07master:b0d94324d21d: libfdk-aacdec: Remove unused variable
[12:26:00 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:7e272100c983: Merge commit '6bcd3e05998b41fb3b2c8dcc79e4662854e5208d'
[12:26:01 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:464dc6e09fd0: Merge commit 'b0d94324d21d0c1c4adf1c8c329928afe473b099'
[12:26:36 CEST] <cone-320> ffmpeg 03Vittorio Giovara 07master:9a003fc38833: lavr: Remove unused label
[12:26:37 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:b8346b22942c: Merge commit '9a003fc38833b6b9f08e44d7f1c6539db65840a0'
[12:27:59 CEST] <cone-320> ffmpeg 03Vittorio Giovara 07master:87e5d8d78cf0: ismindex: Use the correct abs() version
[12:28:00 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:84cca0644a61: Merge commit '87e5d8d78cf08b54b4a9e7cbaeff89f8c1d91b78'
[12:28:47 CEST] <cone-320> ffmpeg 03Vittorio Giovara 07master:a686e58165ca: des: Add missing include
[12:28:48 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:b7601b5f3d16: Merge commit 'a686e58165ca0f83966431a9166cb6e17bf6095c'
[12:30:06 CEST] <cone-320> ffmpeg 03Michael Niedermayer 07master:87a11db220de: dnxhddata: Double frame_size for cid 1260
[12:30:07 CEST] <cone-320> ffmpeg 03Luca Barbato 07master:1a164e0a83dd: tools: Drop a stale header
[12:30:08 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:c8c9940f4745: Merge commit '87a11db220dec523e39eb4af715175e150787a72'
[12:30:09 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:a92bd1901fc0: Merge commit '1a164e0a83dd7c1b129aeef10cb3caff368ce850'
[12:31:42 CEST] <cone-320> ffmpeg 03Luca Barbato 07master:4dc9f0036585: build: Skip libvpx.h if libvpx is not enabled
[12:31:43 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:d4f63be92b4a: Merge commit '4dc9f00365853b1239936d771856cf1881b5c5ce'
[13:06:43 CEST] <cone-320> ffmpeg 03Luca Barbato 07master:0c7707610f46: lavc: Add a lzf decompressor
[13:06:44 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:1964aeae42fd: Merge commit '0c7707610f46a28b65eff935500c78c9781f8f4f'
[13:31:46 CEST] <cone-320> ffmpeg 03Zhang Rui 07master:b821aed70629: avformat/async: fix integer conversion warning
[13:40:23 CEST] <cone-320> ffmpeg 03Paul B Mahol 07master:09812e120099: avfilter/vf_stereo3d: add checkerboard output format
[13:40:24 CEST] <cone-320> ffmpeg 03Paul B Mahol 07master:4b2bafbd4c8e: avfilter/vf_stereo3d: support interleaved rows as input format
[13:40:25 CEST] <cone-320> ffmpeg 03Paul B Mahol 07master:8bd13eb8be59: avfilter/vf_stereo3d: add HDMI output format
[13:40:26 CEST] <cone-320> ffmpeg 03Paul B Mahol 07master:32dd9b2c6e95: avfilter/vf_stereo3d: add interleave columns output format
[13:40:27 CEST] <cone-320> ffmpeg 03Paul B Mahol 07master:0c511eb272e0: avfilter/vf_stereo3d: stop dumping half of data for interleave rows output
[13:44:54 CEST] <cone-320> ffmpeg 03Rostislav Pehlivanov 07master:b6cc8ec7ecc3: aaccoder: add frequency scaling and quantization correction for PNS
[13:46:15 CEST] <cone-320> ffmpeg 03Paul B Mahol 07master:50d9121afc66: doc/filters:vf_stereo3d: update to recent changes
[14:01:14 CEST] <BBB> so regarding this whole abi compat hack thing
[14:01:24 CEST] <BBB> do we have a decision? do we even have a decision making process?
[14:03:31 CEST] <atomnuker> what abi compactibility? haven't really been keeping up with the news
[14:08:03 CEST] <BtbN> There is an IRC meeting scheduled to discuss it.
[14:08:04 CEST] <ubitux> BBB: can it wait til saturday?
[14:10:48 CEST] <atomnuker> ah you mean the Libav ABI compat
[14:33:17 CEST] <BBB> atomnuker: yes
[14:33:38 CEST] <BBB> ubitux: I doubt well reach any sort of consensus on saturday
[14:34:28 CEST] <ubitux> well we can most likely decide how to move on for the change of policy
[14:34:30 CEST] <BBB> ubitux: the discussion has reached a stage where any holdout is taken as evidence for no consensus - i.e. no decision, and even if we get to a stage where theres a clear majority, they invoke aristotle (a fool can ask more questions than many wise men can answer) to delay decision making
[14:34:55 CEST] <ubitux> that's what we want to discuss
[14:34:59 CEST] <atomnuker> I'll bring popcorn
[14:35:05 CEST] <BBB> I love popcorn
[14:35:31 CEST] <BBB> ubitux: ok, fine, well wait until saturday
[14:35:31 CEST] <j-b> salty or sugary?
[14:35:32 CEST] <durandal_170> just push patch
[14:35:34 CEST] <ubitux> BBB: if you want to move fast on this one, my suggestion would be to send a patch for a website news to announce the change of policy (or whatever other mean)
[14:36:14 CEST] <ubitux> then you can move on with the push of the patchset
[14:36:47 CEST] <ubitux> durandal_170: i think most ppl agree with the patch itself
[14:37:02 CEST] <ubitux> the discussion is about aknowledging the change of policy on that regard
[14:37:21 CEST] <ubitux> you can't do that on a whim
[14:37:28 CEST] <ubitux> (remember 2011?)
[14:37:51 CEST] <ubitux> it's not very wise to overreact on such issue
[14:48:50 CEST] <BBB> I wasnt going to do anything
[14:48:55 CEST] <BBB> Im not interested in politics
[14:49:03 CEST] <BBB> Im interested in video codecs
[14:49:29 CEST] <j-b> I'm interested in politics, I'm not interested in video codecs :)
[14:51:08 CEST] <BBB> I hereby give my proxy vote to j-b
[14:51:13 CEST] Action: BBB goes to office
[14:51:14 CEST] <BBB> bbl
[15:32:38 CEST] <cone-320> ffmpeg 03Gwenole Beauchesne 07master:aea611dc3e9e: vaapi: fix local header include.
[15:34:03 CEST] <cone-320> ffmpeg 03Gwenole Beauchesne 07release/2.8:eaabfe8ef8e2: vaapi: fix local header include.
[15:44:42 CEST] <nevcairiel> why is avcodec_open2 being so evil and overwriting width/height based on coded_width/height
[15:47:27 CEST] <nevcairiel> so this particular codec sets coded_width in its init function by aligning widh to 16, since it needs some extra space, apparently. this is during find_stream_info .. the second time avcodec_open2 is called, for the actual decoding, coded_width is already aligned, and avcodec_open2 will set width to the aligned coded_width value, and everything is broken
[15:47:30 CEST] <nevcairiel> hooray for evil hacks
[15:49:40 CEST] <BtbN> The libva-intel-driver commitet strange stuff yesterday, which seems to be related to VP9 decoding.
[15:50:04 CEST] <BtbN> It looks like they added the ability for the intel-driver to chain-load another libva driver, and call it for certain codecs.
[15:54:19 CEST] <nevcairiel> guess i'll just extent the list of codec exceptions by one and call it a day
[15:59:11 CEST] <BBB> BtbN: this has nothing to do with ffvp9 right?
[15:59:30 CEST] <nevcairiel> it might, if he decides t oadd vaapi support for vp9
[15:59:38 CEST] <BBB> hm...
[15:59:45 CEST] <BBB> I meant I dont have to review code"
[15:59:53 CEST] <nevcairiel> not yet :p
[16:00:10 CEST] <BBB> good
[16:00:13 CEST] <BBB> :)
[16:08:20 CEST] <cone-320> ffmpeg 03Vittorio Giovara 07master:c45fcf30cfab: DXV decoder
[16:08:21 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:d02d208211eb: Merge commit 'c45fcf30cfab687004ed1cdc06ebaa21f4262a0b'
[16:10:17 CEST] <cone-320> ffmpeg 03Luca Barbato 07master:5a1a9da8a7ae: mpegvideo: Drop a stray error message
[16:10:18 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:1789e46d34c1: Merge commit '5a1a9da8a7ae120f2543b8f2fa13dc8baac39f17'
[16:15:26 CEST] <cone-320> ffmpeg 03Sean McGovern 07master:e05f7ed54362: file: properly forward errors from file_read() and file_write()
[16:15:27 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:88e7ea3e5603: Merge commit 'e05f7ed5436207f4a55f1978b223c7f8bc82af42'
[16:22:06 CEST] <cone-320> ffmpeg 03Sean McGovern 07master:9ad1e0c12caa: mux: Make sure that the data is actually written
[16:22:06 CEST] <cone-320> ffmpeg 03Hendrik Leppkes 07master:1f1c10083352: Merge commit '9ad1e0c12caa440de860bd8f2122beb11d73815c'
[16:29:16 CEST] <ubitux> durandal_170: just put "Copyright (c) 2015 The FFmpeg Project" if carl is too annoying
[16:29:27 CEST] <ubitux> (we have a bunch of files following that pattern)
[16:31:11 CEST] <durandal_170> i wonder should it really clip, better leave that to another filter
[16:36:20 CEST] <cone-320> ffmpeg 03Ganesh Ajjanagadde 07master:3e830b6dc844: configure: colorize warning messages
[16:38:47 CEST] <durandal_170> who is our current project leader acording to carl?
[16:40:00 CEST] <iive> There is only one Great Leader!
[16:40:43 CEST] <j-b> and only one ABI
[17:51:56 CEST] <cone-320> ffmpeg 03James Almer 07master:2bef340cb2c3: avcodec: remove duplicate AVCodecIDs
[17:51:58 CEST] <cone-320> ffmpeg 03James Almer 07master:83c6a880a668: avcodec: remove fourcc values from AVCodecIDs
[17:54:31 CEST] Action: Daemon404 looks at this latest abi thing on the ml..
[17:54:32 CEST] <Daemon404> 1/g 31
[17:55:54 CEST] <BBB> 1/g 31?
[17:56:03 CEST] <Daemon404> a typo
[17:56:08 CEST] <Daemon404> one too many '1's
[17:56:21 CEST] <BBB> ah
[18:11:11 CEST] <durandal_170> is there good high bit depth video for testing, 10 or 12 depth?
[18:24:06 CEST] <JEEB> durandal_170: either tears of steel or BBB with their UHD/4K renders
[18:24:18 CEST] <JEEB> too bad BBB is PITA to get as master
[18:24:26 CEST] <JEEB> so I guess tears of steel is simpler
[18:24:33 CEST] <BBB> ?
[18:24:34 CEST] <Daemon404> i thought xiph hosted it
[18:24:40 CEST] <Daemon404> BBB, big buck bunny
[18:24:40 CEST] <BBB> oh
[18:24:54 CEST] <BBB> we need a new TLA for that video
[18:24:55 CEST] <JEEB> Daemon404: yeah but not the 2160p60 render master
[18:25:12 CEST] <JEEB> you can only get it as the blender output files
[18:25:16 CEST] <JEEB> scene per scene or so
[18:25:19 CEST] <JEEB> shit sucks
[18:26:45 CEST] <durandal_170> UHD is too much for this dead horse, any smaller resolution?
[18:27:20 CEST] <JEEB> just downscale?
[18:28:22 CEST] <JEEB> the slowest thing with ToS masters is with the fact that it's a JPEG2000 master
[18:30:04 CEST] <BtbN> I have a big buck bunny 4K60 file here. Is that just upscaled?
[18:30:16 CEST] <j-b> or re-rendered?
[18:30:27 CEST] <BtbN> Does that matter?
[18:30:44 CEST] <j-b> yes
[18:31:06 CEST] <JEEB> BtbN: it's a re-render
[18:31:27 CEST] <JEEB> the issue is that IIRC there was no lossless master version available, just lossy re-encodes :/
[18:31:47 CEST] <JEEB> which is why everyone seems to be using the 2160p60 encode as the source that ends up looking crap
[18:32:14 CEST] <Daemon404> there are 4k exrs
[18:32:18 CEST] <Daemon404> 3d even iirc.
[18:32:26 CEST] <JEEB> yeah
[18:32:40 CEST] <JEEB> it was rendered both 2160p60 as well as 2160p60 multiview
[18:33:00 CEST] <JEEB> (inb4 koda)
[18:48:16 CEST] <Daemon404> IT LIVES
[18:48:50 CEST] <atomnuker> you can't kill it now, it's too big
[18:50:33 CEST] <BtbN> atomnuker, btw., regarding that "nice to know someone is using it": The OBS Multi-Platform version uses the ffmpeg aac encoder by default.
[18:51:10 CEST] <BtbN> So a lot of live streams are using it.
[18:51:20 CEST] <Daemon404> yeah but OBS probably doesnt care about audio much...
[18:51:36 CEST] <Daemon404> quality of a guy yelling at some LoL/DotA players probably isnt thyat big of a deal
[18:51:42 CEST] <Daemon404> <_<
[18:51:43 CEST] <atomnuker> I thought as much, I remember you had to download ffmpeg for some
[18:51:51 CEST] <nevcairiel> my $work uses the aac encoder when live encoding for streaming to mobile
[18:52:14 CEST] <nevcairiel> so a bunch of people probably use it daily when streaming videos to their phones from our software
[18:52:21 CEST] <nevcairiel> I do for sure
[19:42:48 CEST] <philipl> JEEB: weird. They've got lossless full resolution frame-by-frame files but only for the 3d version.
[19:43:14 CEST] <Daemon404> yep
[19:43:37 CEST] <Daemon404> because there was such a high demand.
[19:44:08 CEST] <philipl> Can't get enough of that three-dee.
[20:28:03 CEST] <nevcairiel> cant you just take one eye and done
[20:28:03 CEST] <nevcairiel> :D
[20:30:47 CEST] <cone-320> ffmpeg 03Michael Niedermayer 07master:9bd68dea2b7f: avformat/http: Add reconnect_at_eof and reconnect_streamed options
[20:30:48 CEST] <cone-320> ffmpeg 03Michael Niedermayer 07master:bdbab4a35159: avformat/http: exponentially delay reconnection to avoid flooding the server
[20:55:10 CEST] <jamrial> nevcairiel: i'm not sure if the deprecated vima decoder should be an alias
[20:55:23 CEST] <nevcairiel> its the same decoder
[20:55:33 CEST] <jamrial> it has its own entry in allcodecs.c and codec_desc.c, which i forgot to remove
[20:56:00 CEST] <nevcairiel> but it had the same codec id before the bump as well
[20:56:14 CEST] <nevcairiel> and i dont really know what the rules for the d ecoder are
[20:56:16 CEST] <nevcairiel> I wanted to remove it
[20:56:21 CEST] <nevcairiel> but someone might request it by name
[20:56:24 CEST] <jamrial> i know it's the same, but turning it into an alias and removing the stuff from avcodec.c and codec_desc.c will make -c:v vima fail
[20:56:28 CEST] <nevcairiel> is that considered part of the stable ABI
[20:56:39 CEST] <nevcairiel> it always was an alias
[20:56:52 CEST] <nevcairiel> just not an obvious one
[20:57:18 CEST] <jamrial> ok, should i remove the stuff from those two files and leave just the alias in avcodec.h then?
[20:57:28 CEST] <durandal_170> remove it, its game codec
[20:57:40 CEST] <Daemon404> lol
[20:57:49 CEST] <nevcairiel> i suppose its rather unlikely that someone is going to use -c:a vima
[20:57:59 CEST] <nevcairiel> and even if, just let them update to -c:a adpcm_vima :p
[20:58:08 CEST] <jamrial> right, audio
[20:58:10 CEST] <Daemon404> dont tell carl though
[20:58:14 CEST] <jamrial> ok then
[20:58:35 CEST] <nevcairiel> i have no clue if that is considered part of the stable ABI or API now
[20:58:51 CEST] <nevcairiel> seems silly to keep the duped entry
[20:59:07 CEST] <iive> if you can fix the problem with recompilation, it is ABI. if not, it is API
[20:59:26 CEST] <jamrial> well, it's all wrapped by a FF_API deprecation thingy
[20:59:37 CEST] <nevcairiel> the question is how far does that extent :p
[20:59:43 CEST] <nevcairiel> someone may rely on a particular bug which we fix
[20:59:52 CEST] <nevcairiel> thats clearly not considered API :P
[21:00:44 CEST] <nevcairiel> but this codec is so obscure that i doubt anyone cares
[21:00:50 CEST] <nevcairiel> so just strip everything but the codec id
[21:02:53 CEST] <jamrial> mmh, also in libavcodec/vima.c
[21:03:24 CEST] <nevcairiel> yeah thats the ac tual decoder
[21:03:59 CEST] <nevcairiel> which is just a copy of the adpcm_vima 
[21:13:10 CEST] <cone-320> ffmpeg 03James Almer 07master:f9fd87987734: avcodec: remove missing bits from deprecated VIMA decoder
[21:21:25 CEST] <durandal_170> jamrial: Makefile entry is still there
[21:25:39 CEST] <cone-320> ffmpeg 03James Almer 07master:ef359e724d99: avcodec: remove VIMA Makefile entry
[21:25:47 CEST] <jamrial> durandal_170: done
[21:34:05 CEST] <atomnuker> is there an PRNG for getting 0->1 float random numbers somewhere?
[21:34:17 CEST] <atomnuker> random.h was removed 6 years ago from libavutil
[21:35:39 CEST] <durandal_170> AVLFG
[22:13:57 CEST] <cone-320> ffmpeg 03Ivan Uskov 07master:cc167f7e55da: libavcodec/qsvdec.c: correct handling of dynamic frame size changing has been implemented
[22:13:58 CEST] <cone-320> ffmpeg 03Ivan Uskov 07master:3f8e2e995324: libavcodec/qsvdec.c: correct flush() handler has been implemented
[22:19:39 CEST] <cone-320> ffmpeg 03Ganesh Ajjanagadde 07master:b45ff1386ae2: avfilter/vf_thumbnail: use the name 's' for the pointer to the private context
[23:03:03 CEST] <cone-320> ffmpeg 03hSÇ 07master:f7e9b9d260f9: avcodec: loongson optimize h264dsp idct and loop filter with mmi
[00:00:00 CEST] --- Tue Sep  8 2015