[Ffmpeg-devel-irc] ffmpeg-devel.log.20190824

[burek]

[01:34:29 CEST] <rcombs> is it just me or can avcodec_open2 return while still holding codec_mutex in error cases
[01:34:51 CEST] <rcombs> specifically free_and_end
[02:09:07 CEST] <cone-124> ffmpeg 03Aman Gupta 07master:c62a1db0ac16: MAINTAINERS: add myself to OMX
[02:09:08 CEST] <cone-124> ffmpeg 03Aman Gupta 07master:76c82843ccad: avcodec/omx: add support for -force_key_frames
[02:09:09 CEST] <cone-124> ffmpeg 03Aman Gupta 07master:23a3e1460a7a: avcodec/omx: ensure zerocopy mode can be disabled on rpi builds
[02:09:10 CEST] <cone-124> ffmpeg 03Dave Stevenson 07master:3d857f219eb9: avcodec/omx: Fix handling of fragmented buffers
[02:12:09 CEST] <cone-124> ffmpeg 03Aman Gupta 07release/4.2:e008f89cfa1a: avcodec/omx: ensure zerocopy mode can be disabled on rpi builds
[02:12:10 CEST] <cone-124> ffmpeg 03Dave Stevenson 07release/4.2:3dd3e8e24ac9: avcodec/omx: Fix handling of fragmented buffers
[07:54:52 CEST] <cone-195> ffmpeg 03Pavel Koshevoy 07master:22a14ee753f3: avcodec/videotoolbox_hevc: avoid leaking cached_hw_frames_ctx
[09:34:16 CEST] <cone-195> ffmpeg 03Thilo Borgmann 07master:1cfba7fc7584: fate/als: Add test for conformance file with 512 channels.
[16:25:33 CEST] <durandal_1707> looks like nobody except thilo and me wants meeting
[16:26:18 CEST] <JEEB> what sort of meeting was it about? the thread seemed long so I didn't actually read through it
[16:26:27 CEST] <JEEB> not that I'm a new developer :P
[16:29:07 CEST] <durandal_1707> its not about new developers
[16:30:08 CEST] <durandal_1707> its meeting where we pick new leader. seriously there is bunch of stuff to decide/do
[16:31:04 CEST] <JEEB> ok, the thread title or description that I read in gmail seemed to have something about new developers and it seemed to get long'ish
[16:31:30 CEST] <durandal_1707> lol
[16:33:13 CEST] <durandal_1707> we have no new devs, we only lose old and new ones per day :(
[16:34:03 CEST] <JEEB> well I'm pretty sure we've gained people in our git history during the last year
[16:34:13 CEST] <JEEB> of course how active those people become is a different discussion
[16:48:48 CEST] <durandal_1707> modern people not interested in reading long threads
[16:49:51 CEST] <JEEB> esp. if it doesn't seem to be relevant to me? which you thankfully have noted to be an incorrect assumption
[16:50:01 CEST] <JEEB> also nothing modern
[16:50:14 CEST] <JEEB> just priorities of where I can or want to put my time into
[16:50:52 CEST] <durandal_1707> yea, old story
[17:55:22 CEST] <kierank> Meetings are pointless we did it during fork days
[17:55:27 CEST] <kierank> Nobody turned up
[17:55:41 CEST] <kierank> Was blocked by mplayer people
[17:56:28 CEST] <thilo> how to block a meeting?
[17:56:53 CEST] <thilo> whatever, jest reply to the thread and more people will notice if it stays somewhere on top the latest msgs
[17:58:33 CEST] <thilo> if there will be enough people in favor for that and actually enter there dates of comfort there will be another registration thread anyway - and then even the most lazy should notice
[17:58:44 CEST] <thilo> *their
[17:59:36 CEST] <thilo> I just wonder the people in favor that already replied to durandals original thread do not reply to the proposal
[17:59:56 CEST] <durandal_1707> but where you plan to put actual meeting, on open air?
[18:00:13 CEST] <durandal_1707> these need more planning
[18:00:27 CEST] <durandal_1707> no rushing needed
[18:00:38 CEST] <thilo> conference hotel. we don't need anything else but a room big enough for everyone
[18:01:13 CEST] <thilo> no rush needed
[18:07:10 CEST] <kierank> They blocked the actions of the meeting
[18:07:28 CEST] <kierank> We had so many "fork room" meetings it was boring
[18:10:02 CEST] <thilo> have these been like the "fork talks" at later VDDs I witnessed or dedicated meetings just for that purpose?
[18:14:54 CEST] <kierank> Fork talks mainly
[18:15:53 CEST] <thilo> if they were not dedicated meetings, it would be easy to block because just a bunch of people met somewhere talking something
[18:16:22 CEST] <thilo> this time, if enough people want to attend, it will be an announces dev meeting for the project. including decisions
[18:16:47 CEST] <thilo> wether or not michaelni is there or not
[18:17:41 CEST] <kierank> I don't see how this will be different, the usual suspects (sorry to use that terminology but there is nothing better) will just block or filibuster as they have been used to for decades
[18:18:24 CEST] <kierank> I admire your enthusiasm however
[18:18:34 CEST] <kierank> But personally I have seen progress dashed time and time again
[18:19:59 CEST] <thilo> we all have. i've never seen an announced dedicated meeting, though. just "fork talks" at VDD
[18:20:07 CEST] <thilo> has there ever been such one?
[18:22:32 CEST] <kierank> No, but this was out of convenience when many developers were in the same place (e.g fosdem)
[18:23:08 CEST] <durandal_1707> kierank: apply two dhav patches i just posted if you are bored
[18:23:27 CEST] <kierank> Can't, not got laptop
[18:24:02 CEST] <kierank> What is dhav
[18:24:06 CEST] <BBB> thilo: I organized one 3 years ago I think at VDD
[18:24:10 CEST] <BBB> the one in Germany
[18:24:18 CEST] <thilo> well then whatever they agreed on is easy to block
[18:24:36 CEST] <BBB> we typically have more informal group get-togethers at VDD but not to discuss policy, just to listen to users
[18:24:44 CEST] <durandal_1707> dhav is dahua cctv container
[18:25:48 CEST] <thilo> BBB: I've been there... this is exactly what I mean - had happened often but could be blocked for being "non-official" easily
[18:26:24 CEST] <kierank> What will happen is someone will say "there isn't consensus" based on some arbitrary decision
[18:26:37 CEST] <kierank> And then nothing changes
[18:26:39 CEST] <thilo> this time we can just make it "official" and announce it on the ML. And since the ML is considered the "official" channel of communication, hard to neglet legitimacy of that
[18:27:03 CEST] <kierank> Might work
[18:27:20 CEST] <kierank> A step in the right direction I guess
[18:27:36 CEST] <thilo> well, finding consensus is a task for all attendees of course
[18:28:07 CEST] <kierank> Yes but the people who do not travel say "oh but we should have meeting on IRC or ML or whatever, not official decision, blah blah"
[18:28:26 CEST] <kierank> There is no real way to enforce these decisions
[18:29:22 CEST] <thilo> yes, at least it would be moving forward and stop being stalled
[18:29:50 CEST] <BBB> I think we should try to be proactive with allowing non-attendees to participate also
[18:29:57 CEST] <BBB> can we do a live-type out of the discussion?
[18:30:00 CEST] <BBB> maybe live-stream it?
[18:30:05 CEST] <BBB> and then have irc participation?
[18:30:10 CEST] <BBB> (for questions/comments)
[18:30:13 CEST] <thilo> that I can very well imagine
[18:30:26 CEST] <BBB> I know some ML people hate IRC but you can't have everything...
[18:31:28 CEST] <thilo> we might easily screen a freenode channel and stream "live-ish", I guess... that's kierank's job ;)
[18:33:04 CEST] <thilo> however most important for me is people actually going there
[18:58:19 CEST] <BBB> thilo: I think VDD is your best shot
[18:58:30 CEST] <BBB> in other cases you will only have a fraction of people attending
[18:59:39 CEST] <durandal_1707> vdd in tokyo!?
[18:59:52 CEST] <nevcairiel> Pick a VDD thats closer to the core developers though, and not  in Tokyo, really not a fan of 20 hour flights =P
[19:09:34 CEST] <durandal_1707> you closed nicolas report
[19:10:41 CEST] <durandal_1707> how rude :p
[20:38:47 CEST] <durandal_1707> michaelni: you are very rude
[20:39:18 CEST] <durandal_1707> still sending patches that break decoders
[20:39:44 CEST] <durandal_1707> and not providing speedup numbers at all
[20:40:13 CEST] <durandal_1707> why you care for that decoder at all?
[20:40:59 CEST] <durandal_1707> making vfr from cfr is unacceptable
[20:42:01 CEST] <durandal_1707> thinking about not developing new decoders at all because of your devastating changes
[20:43:29 CEST] <durandal_1707> fraps, qtrle, scpr and others too are all victims of your changes
[20:44:18 CEST] <durandal_1707> you forgot to ruin msrle 
[20:44:37 CEST] <thilo> BBB: that would just end up like the examples kierank told about
[20:45:54 CEST] <thilo> BBB: we are overdue about having our own, and be it for the lessons of prior VDD talks alone
[20:47:04 CEST] <thilo> BBB: we might attach it before or after VDD and share resources, that would be good. but we need something else than one single "FFmpeg" session at VDD
[20:47:49 CEST] <jamrial> durandal_1707: are you really against a flag or option to choose between the two modes?
[20:48:36 CEST] <durandal_1707> is that covering decoder compatibility?
[20:49:07 CEST] <durandal_1707> its about default behavior
[20:49:45 CEST] <durandal_1707> and what makes decoder compliant with reference implementation
[20:51:37 CEST] <jamrial> we can make crf the default if the decoder is expected to output that to be complaint
[20:52:36 CEST] <jamrial> i suggested to map it to -flags2 fast but michaelni_ argued it was not ideal
[20:53:09 CEST] <jamrial> i don't really agree because if you care about bitexactness, you use -flags bitexact
[20:55:05 CEST] <durandal_1707> i expect from him to show actual numbers
[20:58:16 CEST] <BBB> thilo: I like the idea, but nevcairiel's comment about vdd at tokyo being far away is true for all events
[20:58:27 CEST] <BBB> thilo: if you do it in croatia, mind you that's far away for me
[20:58:36 CEST] <BBB> I don't like flying over just for a one-day or half-day ffmpeg event
[20:58:40 CEST] <BBB> and that's not just me :-/
[20:58:55 CEST] <BBB> so ...
[20:58:58 CEST] <BBB> it's difficult :)
[21:02:22 CEST] <thilo> sure. I like the idea of doing it a day before or after VDD, also for that reason. For this year VDD... well its Tokyo and less people will come anyway
[21:04:03 CEST] <thilo> and i guess a day extra just for the FFmpeg people might not be what JB had in mind in his ML post about that
[21:05:19 CEST] <thilo> but I can ask him
[21:05:59 CEST] <thilo> however, I also want durandal there... then, Tokyo is not an option
[21:21:08 CEST] <michaelni_> jamrial, a video with and without bitexact flag looks ok, a video decoded with flags2 fast doesnt neccessarily look ok. 
[21:24:54 CEST] <jamrial> michaelni_: i see
[21:30:35 CEST] <durandal_1707> BBB: what you live in antartica? so tokyo is closer
[21:32:39 CEST] <nevcairiel> Personally I would obviously prefer EU locations, but US would be acceptable, but 20 hours flying is just over the top for me
[21:35:18 CEST] <cone-420> ffmpeg 03Paul B Mahol 07master:96fd07351a8d: avformat/dhav: gracefully handle corrupted files
[21:35:18 CEST] <cone-420> ffmpeg 03Paul B Mahol 07master:45cefca1e799: avformat/dhav: handle files missing data at start
[21:42:28 CEST] <durandal_1707> calling for mov timecode reviewers!
[21:52:41 CEST] <durandal_1707> i remember days when reviewing was honorable action
[22:08:14 CEST] <durandal_1707> but i guess nothing lasts forever
[22:32:13 CEST] <thardin> that doesn't look like timecode
[22:35:33 CEST] <thardin> croatia, eh?
[23:00:38 CEST] <thardin> perhaps I should arrange travel to amsterdam
[00:00:00 CEST] --- Sun Aug 25 2019