wiki:FFmeeting/2020-02

This FFmeeting was hosted at the FOSDEM in Brussels, Belgium at 13 UTC. Its audio was streamed to somewhere and a script posted on irc://irc.freenode.net/ffmpeg-meeting on 2020-02-01.

Meeting Blackboard

Full meeting log

[14:10:59] <j-b> hello
[14:11:01] <j-b> https://hangouts.google.com/call/jYaO0pADYZELBBfsntHgAEEI
[14:11:13] <thardin> hullo
[14:11:40] <J_Darnley> I can't invite, need op
[14:13:07] <thardin> ugh google wants me phone#
[14:13:12] <thardin> my
[14:14:23] <Lynne> just use talky.io
[14:14:47] <JEEB> I hope I'm showing up as muted since this UI isn't making me sure if I am or not (I should be)
[14:15:03] <j-b> Do you people hear us?
[14:15:20] <JEEB> no audio so far
[14:15:34] <jamrial> no
[14:15:35] <StevenLiu> No 
[14:15:51] <wbs> I'm just following irc, not the hangout unfortunately
[14:16:10] <JEEB> ok, james's video feed picked up
[14:16:24] <j-b> JEEB: with sound ?
[14:16:42] <J_Darnley> neat
[14:16:56] <JEEB> no sound still but I can just attempt to re-join
[14:17:19] <JEEB> nope
[14:17:27] <JEEB> ok, audio
[14:17:28] <StevenLiu> yes
[14:17:31] <jamrial> yeah
[14:17:32] <StevenLiu> have audio
[14:20:13] <thardin> I'm in. idling with mic off
[14:26:55] <thardin> usually what you do is have a nomination committee that asks people in advance and then present the nominees
[14:27:53] <cehoyos> Can everybody hear?
[14:28:14] <thardin> I can hear
[14:28:20] <JEEB> voting 1: 3d, vote 2: a week, so seems like the conn is working here :)
[14:28:21] <StevenLiu> I can too
[14:28:23] <cehoyos> Atm we don’t copy into irc what is said
[14:29:08] <JEEB> (v1 was IIRC people nominated who might not otherwise show up on voting list, v2 was committees, right?)
[14:29:15] <Illya> git log --since="last 36 months" --author="name" --oneline | wc -l
[14:29:16] <Illya> yes
[14:29:18] <cehoyos> Jeeb: Please write short summaries about what you hear
[14:29:24] <BBB> the hangout in the topic is empty btw
[14:29:31] <cehoyos> (mobile phone here)
[14:29:36] <JEEB> BBB: https://hangouts.google.com/call/jYaO0pADYZELBBfsntHgAEEI
[14:30:16] <JEEB> cehoyos: will attempt.
[14:30:22] <StevenLiu> git log --no-merges  --since=2020-01-25T00:00:00Z --until 2020-02-01T00:00:00Z --pretty=fuller | grep '^Author:' | sed 's/<.*//' |sort | uniq -c | sort -nr
[14:31:03] <cehoyos> Ty
[14:31:06] <JEEB> j-b noting - CoC more like a values list as opposed to specific rules. there will be a suggestion which would then be voted on
[14:33:08] <JEEB> Lynne noting - various audio decoders do checks already done avcodec common utils
[14:33:17] <JEEB> (if I acught that right)
[14:33:41] <michaelni> i have some difficulty understanding lynne with my headphones
[14:35:24] <Lynne> michaelni: the sample rate and other checks in audio decoders that are now checked internally by the API so they should be removed
[14:35:39] <Lynne> you added them, I pinged you on IRC and you didn't remove them
[14:36:07] <michaelni> Lynne, i dont remember abouzt the ping but yes if there are redundant checks i should remove them
[14:36:15] <michaelni> ping me again until i react!
[14:36:42] <JEEB> for new joiners: since the topic is out of date if you want to join muted the URL is https://hangouts.google.com/call/jYaO0pADYZELBBfsntHgAEEI
[14:36:59] <jamrial> patches would not be "lost" if we move to gitlab, for example
[14:37:32] <JEEB> gitlab move: I guess main part being discussed atm being merge requests
[14:37:44] <thardin> if patches are handled by say gitlab, is it possible to subscribe via rss/atom?
[14:38:01] <JEEB> I think yes, you can cehck with videolan's gitlab instance
[14:38:45] <JEEB> couldn't find RSS/atom right away, but they have JSON https://code.videolan.org/videolan/x264/merge_requests.json
[14:38:50] <thardin> ugh
[14:38:56] <JEEB> (just giving x264 as an example)
[14:39:02] <thardin> I keep track of mxf issues over rss
[14:39:11] <thardin> which is really handy
[14:39:21] <haasn> thardin: there are atom feeds for project activity, not sure if there's one *specific* to MRs
[14:39:27] <JEEB> ah
[14:39:33] <thardin> haasn: that might be enough
[14:39:46] <thardin> rss readers typically haev filters
[14:39:57] <michaelni> i dont see the problem with the existing infrastructure, so i dont see why we should move to gitlab
[14:40:05] <haasn> e.g. https://code.videolan.org/videolan/dav1d.atom
[14:40:47] <thardin> I run a gitlab instance at uni, and one thing I've found with gitlab is that it's.. a big thing. like it sometimes breaks for seemingly random reasons
[14:42:16] <JEEB> yes, it's a very large ruby on rails thing, which is why I would hopefully share the system with another project, like videolan
[14:42:31] <thardin> that sounds like a good idea
[14:43:01] <thardin> I upgraded our instance when the last ubuntu lts came out, which was a bit of a chore but now I don't have to care about it for a long time
[14:43:01] <haasn> (one consideration that shouldn't be ignored IMHO is that MRs also have a lower barrier for entry for outside contributions, since adopting a ML workflow for people not familiar with the project is a nonzero amount of hassle)
[14:43:45] <michaelni> Lynne, when you have time please send me a mail or something about the checks with fuzzing that you know are redundant and that i should remove
[14:44:31] <thardin> one annoying thing with gitlab is that it requires js just to read tickets
[14:45:43] <JEEB> carl noting that amd and nvidia people nowadays tend to try and be more thoughtful of current FFmpeg design, while Intel seems to attempt to push more driver-specificness there? (if I got it right?)
[14:45:53] <JEEB> (mostly wrt pix_fmts?)
[14:47:32] <JEEB> j-b suggestion: there will be a call to see their reasonings
[14:48:33] <thardin> https://live.fosdem.org/watch/h3242  might also work for people who want to listen
[14:49:25] *** Parts: wm4 (~wm4@p200300E7FF1FD095E15E376A87A246FC.dip0.t-ipconnect.de) ("Leaving")
[14:49:30] <JEEB> (so I guess this is something post-P010?)
[14:49:54] <michaelni> "https://live.fosdem.org/watch/h3242" says "Whoops, looks like something went wrong." for me
[14:51:32] <thardin> yeah, for me too
[14:51:45] <thardin> in the future we should try to use some free protocol, say xmpp
[14:51:59] <JEEB> j-b noting: nvidia was not aware of issues with their headers, there will be discussions wrt that
[14:52:01] <thardin> or H.323 :]
[14:52:19] <cehoyos> We are not aware either...
[14:53:30] <JEEB> j-b noting: should git commit access list have clean-ups?
[14:55:00] <thardin> if we switch to something like gitlab then we could have gitlab-ci automagically run FATE and merge patchsets in once they've been approved
[14:55:13] <JEEB> at least the FATE part, yes
[14:55:43] <JEEB> (probably not all archs/OSs, but at least on some level)
[14:56:08] <thardin> and most importantly, refuse to merge if things break on some (important) platform
[14:57:30] <thardin> yeah no, cmake is terribl
[14:59:21] <JEEB> Lynne noting that gstreamer had a meson version of our build system, which was gaining interest while the bash-based thing was slow. then the configure script got sped up and interest vained
[14:59:43] <JEEB> kierank noting the lavf I/O and demuxing parts being interchanged
[14:59:46] <JEEB> *entangled
[14:59:52] <JEEB> thank you, that is a better word :)
[15:00:59] <JEEB> j-b noting that due to the playlist/manifest based HTTP formats, it has become less clear cut as before
[15:01:55] <JEEB> next meeting around in 1.5 months?
[15:02:04] <jamrial> sounds good
[15:02:16] <StevenLiu> lgtm
[15:03:20] <thardin> sure
[15:03:35] <JEEB> alright, room being cleared up
[15:03:47] <thardin> what's that whistling?
[15:03:57] <JEEB> probably the door, I remember them being funky or so
[15:04:01] <JEEB> (of the room)
[15:04:04] <BBB> is the a summary of the first 20 minutes or was anything important decided? will there be a summary sent to the list?
[15:04:36] <StevenLiu> JEEB: "that due to the playlist/manifest based HTTP formats" which part? demuxer and muxwer?
[15:04:40] <StevenLiu> muxer?
[15:04:48] <JEEB> StevenLiu: that was demuxer side which j-b mentioned
[15:05:13] <StevenLiu> Thanks JEEB and j-b
[15:05:44] <JEEB> BBB: I would expect there to be a minutes posted. the two voting thing were: 1. nominating people who possibly wouldn't get picked up by the commit count rule (three-day vote) 2. committee vote that is free, but if you get nominated you can say no
[15:05:50] <JEEB> (1w vote)
[15:07:48] <cehoyos> Was there an answer?
[15:08:13] <StevenLiu> for which question?
[15:08:25] <durandal_1707> is meeting already over ? :(
[15:08:40] <cehoyos> If other people were on the summary of last irc meeting
[15:08:55] <JEEB> other people?
[15:08:57] <cehoyos> no, apparently the room is only needed in one hour
[15:09:21] <cehoyos> there was no question from me?
[15:09:50] <JEEB> your last message was > 15:52 < cehoyos> We are not aware either... (wrt nvidia headers)
[15:09:53] <JEEB> then a disconnect
[15:10:15] <cehoyos> I asked if - spart from A Strasser, Moritz and Lou (I dont know how many commits they have) should be listed to be voters
[15:10:29] <JEEB> ah yes, that was the first one. which was the first, 3d vote?
[15:10:36] <cehoyos> It seems I didn’t send the question, sorry
[15:12:04] <BBB> JEEB: tnx
[15:12:26] <StevenLiu> ffmpeg-devel-irc's archive program maybe have some problem
[15:12:39] <StevenLiu> http://mplayerhq.hu/pipermail/ffmpeg-devel-irc/2020-January/006244.html
[15:13:01] <cehoyos> We now (today) have three names on the list to be voters even if they don’t have the necessary commits
[15:13:02] <StevenLiu> only "[00:00:00 CET] --- Thu Jan 30 2020 " there
[15:13:07] <cehoyos> Yes
[15:14:29] <durandal_1707> can we move to more crucial issues, irrelevant voters stuff is on the list each and every meeting
[15:16:18] <StevenLiu> isn't that end?
[15:16:34] <cehoyos> Kind of, yes
[15:16:49] <StevenLiu> ok :)
[15:16:52] <JEEB> durandal_1707: there were other discussions as well, you just joined later and thus you just saw someone noting something regarding to one of the first parts of the discussion
[15:17:27] <cehoyos> So my question was: Is there anybody who see last time’s summary and can check if anybody else was mentioned back then?
[15:18:05] <JEEB> right
[15:19:37] <JEEB> last time  was a bit special since it was a boot-up
[15:20:03] <cehoyos> I thought Ronald send a summary 
[15:20:12] <JEEB> yea, I just don't notice a list of special nominees
[15:20:18] <michaelni> StevenLiu, about the IRC log, forwarded your coment to burek
[15:20:19] <JEEB> in > FFmpeg developer meeting 2019/12/9 notes
[15:20:36] <J_Darnley> We're done.
[15:20:40] <JEEB> also thanks for everyone wrt the meeting
[15:20:49] <J_Darnley> I hope you enjoyed the show (if you watched).
[15:20:58] <JEEB> also was nice listening in on the Y210 discussion
[15:20:58] <J_Darnley> I'll have to read this log later
[15:21:12] <durandal_1707> was show recorded in any way?
[15:21:18] <J_Darnley> Not on my end.
[15:21:25] * michaelni had quite some difficulty understanding people through the microphone used today
[15:21:34] <JEEB> anyways, wrt pixel formats, while it is C++ I have recently more and more moved towards zimg
[15:21:49] <durandal_1707> ban him ^
[15:22:05] <JEEB> just noted since someone (j-b?) called for a new pixel format conversion thing on CPUs
[15:22:10] <J_Darnley> Sorry, my shitty internal one
[15:22:47] <J_Darnley> Perhaps I needed to leave it louder.
[15:22:58] <michaelni> no need to be sorry, it was better than nothing :)
[15:22:59] <durandal_1707> called, but with no money provided, they expect people work for free or for miserable paycheck
[15:23:10] <michaelni> dont think louder would have helped 
[15:23:32] <michaelni> also seemed the connection at least the one i had was going up and down to rather low bitrates
[15:24:06] <J_Darnley> kierank for one complains that the wifi here is shit.
[15:24:19] <JEEB> yea, that probably caused the drops at some points which were short thankfully
[15:25:08] <durandal_1707> ffmpeg meeting - with bad infrastructure
[15:25:48] <J_Darnley> I should build my own using a €2000 camera
[15:26:06] <J_Darnley> How do I get Hangouts to ingest it though.
[15:29:43] <cehoyos> Good bye everybody!
[15:34:04] <durandal_1707> so what was discussed if at all? what where conclusions? if any.
[15:38:35] *** Parts: Lynne (~lynne@pars.ee) ()
[16:00:43] <kierank> durandal_1707: will be send by email iirc
[16:12:02] <kierank> j-b took photo
Last modified 4 years ago Last modified on Feb 22, 2020, 7:50:26 PM

Attachments (2)

Note: See TracWiki for help on using the wiki.