wiki:FFmeeting/2020-12

Recording: ā€‹https://youtu.be/1EjIdYuWXEM

--- Log opened Mon Dec 07 18:03:30 2020
Lastlog:
13:17:26 -!- Topic for #ffmpeg-meeting: https://hangouts.google.com/call/xKeJl8WHuMJ67Ue-y0l4AEEM - https://ffmpeg.org/pipermail/ffmpeg-devel/2019-November/252607.html
13:17:26 -!- Topic set by kierank [sid5955@gateway/web/irccloud.com/x-yvyadwozjsmterts] [Sun Nov 10 06:14:53 2019]
13:17:26 [Users #ffmpeg-meeting]
13:17:26 [ BradleyS] [ Illya    ] [ jessidhia] [ koda  ] [ michaelni ] [ thardin] 
13:17:26 [ elenril ] [ J_Darnley] [ jkqxz    ] [ kurosu] [ mindfreeze] [ thilo  ] 
13:17:26 [ ePirat  ] [ JEEB     ] [ kierank  ] [ linjie] [ Nightrose ] 
13:17:26 -!- Irssi: #ffmpeg-meeting: Total of 17 nicks [0 ops, 0 halfops, 0 voices, 17 normal]
13:17:31 -!- Channel #ffmpeg-meeting created Sun Nov 10 06:06:37 2019
13:17:41 -!- Irssi: Join to #ffmpeg-meeting was synced in 22 secs
13:31:27 -!- j-b [~jb@videolan/developer/j-b] has joined #ffmpeg-meeting
14:04:10 -!- j-b changed the topic of #ffmpeg-meeting to: NewTopic
14:04:20 -!- j-b changed the topic of #ffmpeg-meeting to: Dec 5th,  15:00 UTC (7:00 US West, 16:00
14:05:26 -!- j-b changed the topic of #ffmpeg-meeting to: Dec 5th, 15:00 UTC (7:00 US West, 16:00 Berlin/London/Paris, 23:00 Peking). https://ffmpeg.org/pipermail/ffmpeg-devel/2020-November/272272.html
14:08:13 -!- j-b changed the topic of #ffmpeg-meeting to: Dec 5th, 15:00 UTC (7:00 US West, 16:00 Berlin/London/Paris, 23:00 Peking). https://ffmpeg.org/pipermail/ffmpeg-devel/2020-November/272272.html - https://meet.google.com/pgi-xngk-tuq
14:08:26 < j-b> https://meet.google.com/pgi-xngk-tuq
14:09:03 < Illya> london isnt in the same timezone as berlin/paris
14:21:56 < Lynne> can we change the URL to a non-google service?
14:22:16 < Lynne> we talked about this yesterday in ffmpeg-devel, and we agreed jitsi is okay
14:23:33 < Illya> Lynne: test https://meet.jit.si/ShutSpousesBackEither please
14:23:50 -!- jamrial [~jamrial@181.23.80.53] has joined #ffmpeg-meeting
14:29:02 -!- j-b changed the topic of #ffmpeg-meeting to: Dec 5th, 15:00 UTC (7:00 US West, 16:00 Berlin/Paris, 23:00 Peking). https://ffmpeg.org/pipermail/ffmpeg-devel/2020-November/272272.html - https://meet.jit.si/ShutSpousesBackEither
14:29:43 < JEEB> &34
14:33:54 < Lynne> Illya: jitsi accepts all links
14:34:00 < Lynne> so https://meet.jit.si/ffmpeg_meeting works, and I'm in it atm
14:34:53 < Lynne> can anyone who's a mod here change the link?
14:38:19 < Lynne> ping?
14:40:23 -!- j-b changed the topic of #ffmpeg-meeting to: Dec 5th, 15:00 UTC (7:00 US West, 16:00 Berlin/Paris, 23:00 Peking). https://ffmpeg.org/pipermail/ffmpeg-devel/2020-November/272272.html - https://meet.jit.si/ffmpeg_meeting
14:49:36 < kierank> Does Jitsi work on mobile?
14:49:50 < j-b> yes
14:50:07 < j-b> Illya: notes taking somewhere?
14:50:26 < Illya> yep will be
14:58:42 < durandal_1707> why my microphone does not work at all, not detected
14:59:12 < JEEB> if it's through a browser, check that the site has the right to access your microphones?
15:00:46 < JEEB> the initial join page apparently worked for me under firefox
15:02:28 < durandal_1707> in first screen it shows green on bottom that microphone is working
15:03:26 < kierank> durandal_1707: use zoom
15:06:07 < durandal_1707> how? this is jit.si
15:07:03 < jamrial> just join without a mic
15:08:21 < durandal_1707> no, i want to talk in deep bassy voice
15:08:24 < Illya> j-b: https://docs.google.com/document/d/1WN2up7TBC8_pyizgfgkk0duzXCfmN7WFtXXzV6RqaI0/edit?usp=sharing
15:08:33 < j-b> Illya: arghghghgh google doc
15:08:48 < Lynne> yup, needs a google account
15:08:52 < Lynne> -_-
15:08:55 < j-b> bad
15:09:09 < j-b> https://mensuel.framapad.org/p/ffmpeg_meeting
15:11:05 < linjie> : D
15:11:38 < liuqi> linjie good evening :D
15:13:34 < linjie> liuqi: Evening!
15:13:58 < durandal_1707> Paul is muted, cant talk
15:14:07 < elenril> JEEB: your volume is low
15:14:46 < liuqi> I only hear noise
15:15:50 < jamrial> JEEB: your mic volume is low
15:16:13 < JEEB> alright, will check
15:16:19 < j-b> durandal_1707: then, please give your objections here
15:16:52 < michaelni> I can review & comment on the CoC if people want and point me to where to look
15:17:00 < j-b> michaelni: great.
15:18:19 < durandal_1707> well if i use audacity it records chrome output
15:20:57 < durandal_1707> lol, perhaps microphone is just diad
15:20:57 < durandal_1707> *dead
15:23:09 < jamrial> regarding both commitees, can we set a limit or penalty for abusing calling them every time someone disagrees with you or seemingly blocks a patch?
15:24:36 < j-b> jamrial: ok
15:30:22 < jamrial> JEEB: gsoc gave us hevc. but that may have been an exception
15:31:19 < elenril> smarter was VERY exceptional
15:31:26 < jamrial> yeah
15:31:26 < elenril> it was basically a miracle
15:32:34 < michaelni> IMHO we should continue participating in GSoC
15:33:06 < Lynne> as long as ther administrative burden isn't too bad for the people doing it
15:33:07 < liuqi> for example abr(Adaptation BitRate ) or SAMPLE-AES (cbcs) for for hls
15:33:27 < Gyan> Continue.
15:33:35 < elenril> at least gsoc is not a burden on people who don't care about it
15:34:19 < liuqi> Continue
15:35:19  * kierank afk for 15
15:36:09 < kierank> yes
15:50:35 < elenril> j-b: DIE DIE DIE
15:52:00 < elenril> all build systems suck, someone should write a good one
15:52:56 < jamrial> i quite like our current build system as is
15:53:11 < Illya> im not a fan of meson
15:53:21 < liuqi> not too.
15:53:24 < elenril> meson sucks IME
15:54:04 < Lynne> "So speaks the lord of terror, and so it is written."
15:54:29 < jamrial> can't IO be moved to lavu?
15:54:56 < Lynne> I'd rather not tbh, lavu is for more generic platofm-independent code
15:55:04 < elenril> it can, but lavu is already a bunch of random things
15:55:12 < elenril> speaking of which: libavhwcontext
15:55:17 < jamrial> hwcontexts are hardly platform independent
15:55:20 < Lynne> yeah, just remembered -_-
15:55:59 < jamrial> well, it is an utilities library. it's meant to be "random" things, in a way :p
15:56:43 < elenril> as long as those things are reasonably small
15:57:04 < elenril> I would think distros hate how lavu now links to every hwaccel lib under the sun
15:57:24 < jamrial> probably
15:57:33 < elenril> s/small/selfcontained
15:59:17 < kierank> yes
15:59:45 < durandal_1707> what was said about nicolas and me?
15:59:54 < j-b> durandal_1707: that we love you :)
15:59:58 < j-b> durandal_1707: well, I do.
16:00:19 < durandal_1707> i rebooted, and got microphone finally working
16:00:33 < JEEB> yup, your sound worked ok
16:00:56 < durandal_1707> so i joined when people mentioned me and "bad" word
16:02:35 < JEEB> I am not sure if you were mentioned, it was just a "too bad" since even if all logs are posted some people will still not be happy with the logs left from the meeting.
16:03:14 < michaelni> Id like to keep the libpostproc code/feature in some form/place useable from luibavfilter
16:03:48 < kierank> we could put it in a different repo and have it optional?
16:04:00 < kierank> as it was in the beginning
16:04:34 < durandal_1707> libpostproc is already external repo, and abandoned, ask derek
16:09:52 < jamrial> other thing to consider: a patch that no one reviews, is it implicitly accepted after a bunch of pings or not? so far, that seems to be how most developers work
16:10:07 < Gyan> Related to Topic 7Many (most?) patches are unreviewed. We should shepherd patches, especially by new / occasional contributors.Key change needed:  avoid leaving patches in limbo or ignoredProposed process (rough):  acknowledging patch(es) on receipt --> identify / designate reviewers --> alert reviewers as required --> ongoing engagement -->
16:10:08 < Gyan> resolution.
16:10:41 < Gyan> Related to Topic 7  Many (most?) patches are unreviewed. We should shepherd patches, especially by new / occasional contributors.  Key change needed:  avoid leaving patches in limbo or ignored  Proposed process (rough):  acknowledging patch(es) on receipt --> identify / designate reviewers --> alert reviewers as required --> ongoing engagement -->
16:10:41 < Gyan> resolution.
16:16:50 < Gyan> It's a self-reinforcing defect: fewer reviews --> fewer new contributors --> fewer reviewers --> fewer reviews
16:17:02 < JEEB> yup
16:25:06 < michaelni> SPI process IIRC is you buy it and ask for reimbursement on ML, if noone object its forwarded to SPI and then its up to SPI , that is IIRC
16:26:05 < elenril> j-b: power9 is actually way more appropriate since it's open
16:26:12 < jamrial> michaelni: so far that was done for travel costs, but i don't recall it done for hardware
16:26:20 < michaelni> and as mentioned above some value items belong to SPI but are hosted / adimed by someone from ffmpeg, this stuff is docuemneted at SPIs website IIRC
16:26:45 < michaelni> jamrial, i suggest someone tries it with something minor to see how it will work
16:28:33 < elenril> it's rather problematic that your first spend your money on it and only then it's actually decided whether you get the money back
16:28:52 < Lynne> and that its owned by SPI
16:29:01 < Lynne> and SPI buys it?
16:29:30 < Lynne> the way I see it, those money are pretty much only usable for travel expenses or conferences, not for/by developers :(
16:29:30 < Gyan> Can't we just ask someone at SPI?
16:29:47 < jamrial> i'm very much in favor of switching to gitlab
16:31:24 < jamrial> gitlab shows differences between commits after a force push in a given MR. it's incredibly useful and much easier to review with huge patches compared to email
16:31:30 < linjie> Iā€™m for merge- request too, easier to track patches and updated histories.
16:32:56 < jamrial> elenril: afaik you can send emails that gitlab automatically converts into an mr
16:33:21 < elenril> jamrial: I know you can do _some_ things with it
16:33:46 < elenril> but not everything
16:33:59 < elenril> e.g. reviewing was not that some time ago
16:34:02 < elenril> *not there
16:34:54 < elenril> and to be clear - it's not that I'm so much in love with email
16:35:19 < elenril> I just hate modern web browsers, the user experience is atrocious
16:36:35 < linjie> Fine from me : D
16:36:35 < JEEB> problem with something !gitlab is that a lot of the integrations or basic features might not be there. at least gitlab is something that can be issue-reported to :)
16:36:51 < JEEB> (and seems to be the most feature-complete thing)
16:37:54 < jamrial> jkqxz: to fully integrate dav1d libavcodec would need to support frame+slice threading. without that, it will be outperformed by the libdav1d wrapper
16:38:05 < elenril> was about to say
16:38:13 < JEEB> I think that is just a case of mapping that somehow, no?
16:38:30 < elenril> no, that needs to be done in the generic code, not inside the decoder
16:38:37 < JEEB> ok
16:38:51 < elenril> and you want to do that for other codecs too
16:38:55 < elenril> like hevc
16:38:56 < JEEB> sure, like hevc
16:38:56 < JEEB> yup
16:45:00 < elenril> j-b: and then we rename ourselves ffstreamer
16:45:10 < Illya> elenril: +1
16:49:03 < JEEB> jitsi worked pretty nicely overall
16:49:07 < j-b> I will reformat the framapad and send it.
16:49:11 < j-b> JEEB: yup
16:49:17 < j-b> JEEB: a lot better than I rememberd
16:54:10 < Lynne> yeah, this went fine
16:54:22 < Lynne> we haven't had a meeting in almost a year
16:54:39 < michaelni> elenril, about SPI payments, i dont think it should be a real problem that equipment is bought first and then reimbursed afterwards. Its only a problem if ever SPI rejects such a reimbursement, also i think i remember reading that SPI has a credit card with which they could pay (cant dfind now where that was written) but that way they could pay directly
16:55:49 < Illya> who owns the equipment too etc
16:56:07 < Illya> what about developers who need to 'own' their equipment like Lynne
16:56:13 < michaelni> still if SPI rejects a payment (which i dont think would happen) one could return the equipment or fflabs could pay it or iam sure we can find another payer
16:56:14 < Illya> a GPU is a bit of a strange thing to be shared
16:57:05 < michaelni> Illya, the ownership IIUC is a legal requiremnt for SPI being a tax exempt non profit thing, why would one need to "own" the equipment ?
16:57:18 < Illya> see comment above
16:57:33 < Illya> how would SPI own a GPU which is used in a developer owned machine
16:57:37 < Illya> how do you even track that etc
16:57:55 < michaelni> Illya, i dont see the problem 
16:58:23 < michaelni> assume my GPU of my notebook was owned by SPI, it wouldnt make a difference would it ?
16:58:54 < michaelni> i mean unless i sell my notebook or inherit it after death or something
16:59:55 < JEEB> basically it just means that if SPI wants it back, they technically can request that.
17:00:40 < JEEB> tracking as such shouldn't be a problem, an XYZ thing has been purchased and sent to PERSON
17:00:41 < michaelni> btw: "Generally, it is expected that individuals will pay the expenses up front and be reimbursed by SPI, but advance payment can be arranged with sufficient advance warning to the Treasurer.  The Treasurer also has a Visa card for online purchases which for some reason need to be made directly by SPI."
17:03:28 < michaelni> JEEB, i dont think they could legally do that as its payed by donations earmarked for FFmpeg but thats a legal question and iam a lifeform not a lawyer
17:07:36 < Illya> https://youtu.be/1EjIdYuWXEM
17:08:54 < durandal_1707> put it on bluray
17:09:35 < Illya> get to listen to durandal_1707's beautiful voice in glorious 4k hdr
17:18:01 < JEEB> durandal_1707: oh you're here :P on ffmpeg-devel there's someone called jafa who has patches for your ac4 dec
17:23:05 < Lynne> they do?
17:47:09 < kierank> I don't understand what notebook gpu has to do with it
21:17:00 -!- durandal_1707 [~computer@77.237.97.78] has quit [Ping timeout: 256 seconds]
21:20:04 -!- durandal_1707 [~computer@77.237.97.78] has joined #ffmpeg-meeting
22:13:01 < koda> thanks for the recording, this is already leagues ahead better than AOM :)
23:32:33 -!- durandal_1707 [~computer@77.237.97.78] has quit [Ping timeout: 244 seconds]
04:40:56 -!- jamrial [~jamrial@181.23.80.53] has quit []
09:00:04 -!- j-b [~jb@videolan/developer/j-b] has quit [Ping timeout: 256 seconds]
09:00:58 -!- j-b [~jb@natalya.videolan.org] has joined #ffmpeg-meeting
End of Lastlog
Last modified 7 months ago Last modified on Apr 23, 2021, 4:21:33 AM
Note: See TracWiki for help on using the wiki.