18:01:32 #startmeeting 18:01:32 Meeting started Thu Jul 30 18:01:32 2020 UTC. The chair is ivodd. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:01:32 Useful Commands: #action #agreed #help #info #idea #link #topic. 18:01:36 #chair pollo 18:01:36 Current chairs: ivodd pollo 18:01:39 lol 18:01:39 wow 18:01:46 ivodd: no, you have to do it now 18:01:47 #topic who's here 18:01:49 o/ 18:01:51 0/ 18:01:57 \o 18:01:57 i'm here! 18:02:11 \o 18:02:21 #topic video stack - jibri 18:02:34 there was quite some work on jibri, but I don't know the current status 18:03:27 the MR is still a WIP, afaiu, paddatrapper needs to change a few things 18:03:33 but 75% of the work is done? 18:03:37 the MR is 90% there 18:03:44 it's been all details recently 18:03:51 do we have a jibri running somewhere already? 18:04:01 it should be on the jitsi VM right now 18:04:08 I'd like to see us get one up on another machine 18:04:21 #info jibri is mostly running, the MR to get it into ansible is 90% there 18:04:34 tumbleweed: does it feed into vocto? 18:04:44 (I haven't checked voctoweb in the last few days) 18:04:53 ivodd: I haven't seen it do that, yet 18:05:01 but I know it's ansibled onto that box 18:05:37 anything else about jibri? 18:06:08 #topic video stack - voctoweb 18:06:26 tumbleweed continues to have fun with this :) 18:06:33 that seems to be working now? 18:06:47 as long as tumbleweed is having fun, it's good for everyone 18:06:52 it has been working since the beginning 18:07:00 that goes for a great many things, actually, not just voctoweb 18:07:05 although I guess it was a bit crashiy while it was constantly poking at voctocore 18:07:42 tumbleweed: is there anything specific we need to get into (except playback, which is later on the agenda)? 18:08:12 so, it's probably approaching the point where it's useful to work out what controls we want to offer 18:08:17 and what automation we'd like 18:08:36 tumbleweed: by controls you mean presets? 18:08:37 (I'm imagining it'd be nice to be able to queue up another video for after the current playback, or a switch to a different source, for example) 18:08:47 pollo: yes 18:08:56 yeah, what UI buttons, what should they do 18:09:05 tumbleweed: I don't think queing is very important 18:09:06 I suggest we make an issue for that, to avoid spending an hour on that now 18:09:12 +1 18:09:14 pollo: queing is very important :) 18:09:30 ivodd: yeah, I would tend to agree 18:09:31 I don't think it's MVP critical - I think we could run the conference with it as is 18:09:38 but it's certainly nice to have 18:09:44 for crisp transitions 18:10:08 +1 to moving this discussion out into issues 18:10:08 tumbleweed: without 'easy' interface, I think it will be hard to train volunteers, but I think we agree on this and we can discuss this outside the meeting 18:10:15 #topic video stack - loop generation 18:10:44 the idea being we'd like to be able to modify the sponsor loop live% 18:10:46 so, I've been meaning to play with things like https://stackoverflow.com/a/54065773/8629 18:10:47 there have been some suggestions to replace the 'static' loop with something more dynamic, but AFAIK, nothing has happened on this 18:10:51 but I can't do everything :P 18:11:07 ivodd: I've been working on it 18:11:14 tumbleweed: did you get something like that working? 18:11:27 olasd was planning to look at casparcg 18:11:32 ivodd: it's not finished yet. there's also some feature creep from following HOPE (they have some great ideas) 18:11:35 but I think he got stalled on GPU support in the VM 18:11:37 highvoltage: with OBS? 18:11:41 pollo: yes 18:11:45 ivodd: no, I haven't tried yet 18:12:20 highvoltage: what are you planning to use? 18:12:26 ivodd: obs in a VM 18:12:37 personal opinion, but I feel adding OBS on top of our already very heavy stack is too much if it's just for the loop 18:13:04 highvoltage: we have an OBS VM you can play on 18:13:09 it doesn't complicate much, it can stream to jitsi so, it's just showing a jitsi stream really 18:13:21 my feeling is that worst case, we can use the 'static' loop, and if something better shows up, we can use it 18:13:23 pollo: it's easy enough to have the other loop fill in if the obs loop dies 18:13:41 yes, but OBS can't really be automated, can it? 18:13:43 also, if it just runs by itself, volunteers don't need to know the detals (which is important) 18:13:51 pollo: it can (but that requires addons) 18:13:53 someone will have to clikedy click that too 18:13:53 well I know how to use OBS so it's more a case of setting up the scenes and sequences, that's going to be better to set up locally, export, and then add to the vm 18:14:12 sounds good 18:14:16 #info highvoltage is trying to set up something based on OBS 18:14:17 ivodd: yes, of course, static loop is the good fallback option here 18:14:19 pollo: it can be automated 18:14:36 #info olasd is planning to look at casparcg 18:14:42 and if it doesn't work out, we also have mdco#2 in november as another shot to try that out again 18:15:17 about the gstreamer stuff: I did some basic tests a few weeks ago, which worked locally, but as soon as you pipe it into vocto, I get issues with time 18:15:18 ivodd: yeah that loop will run by itself until we want to change something, but volunteers wouldn't need to touch it 18:15:20 so I gave up on it 18:15:49 #info we can use the static loop, but if something nicers is ready in time, we can switch to that 18:16:11 anyone volunteers for design? 18:16:22 for the loop? 18:16:25 yes 18:16:26 I can look at it, or have Tammy do so 18:16:36 I'll see if she's interested later 18:16:41 I'll need a sponsors loop anyway so maybe that's a good place to start? 18:16:53 wouter: imo the "easiest" way to produce the loop is to use the ffmpeg commands in our doc 18:16:56 via a bunch of pngs 18:16:56 we'll need sponsor logos if we're going to do that 18:17:20 if only they were already scattered all over the website 18:17:22 pollo: er, we need some animation software 18:17:23 wouter: DLange did say the sponsors on the website were pretty final 18:17:26 the rest is just output 18:17:33 cool 18:17:45 anything more on this? 18:17:51 sure, just saying the file format needs to be specific 18:17:57 or else we can't loop it :) 18:18:04 just saying it doesn't matter for me :) 18:18:04 right, but that can be fixed later 18:18:33 beautiful loop <- +1 (and many +1 from fundraising team, I'm sure) 18:19:01 (we get two more sponsors as it currently looks, should know within a week) 18:19:07 #action wouter to look at generating sponsor loop design 18:19:13 #action wouter to look at generating sponsor loop design 18:19:24 #chair wouter 18:19:24 Current chairs: ivodd pollo wouter 18:19:39 ivodd: er, you don't need to be a chair to #action yourself, but thanks anyway ;) 18:19:50 wouter: oh - TIL 18:20:08 #topic video stack - sreview 18:20:09 what colours should the sponsor loop background be? *ducks* 18:20:35 so I did some work on that, and the basic infra to inject videos and transcode them to a particular profile is ready 18:20:44 people can upload video recordings and slides etc 18:21:01 these get stored on the SReview host (currently vittoria), and then transcoded to the profile 18:21:05 we keep both the source and the transcode 18:21:20 unfortunately, the current profile still crashes vocto, so I need to fiddle with it a bit more 18:21:25 #info sreview is ready, people can upload video recordings and slides etc 18:21:46 but it would probably help if we could figure out *why* that broken video crashes vocto, excactly, so I can create the right profile that avoids that issue 18:22:00 other than that, we're good 18:22:02 at least, with broken inputs it does 18:22:07 well, yes 18:22:14 highvoltage: same colour as the bike shed, obvs 18:22:15 there are presumably things that go through it and don't break vocto 18:22:36 sure, but the whole point of this exercise was so that we come up with video files that don't break vocto 18:22:47 if we go through this and it still happens, that's kindof missing the point :) 18:23:50 anyway, if it's unlikely that this will happen with actual videos that people upload, I'm not bothered, but we don't know why vocto crashes, so I don't want to say that it's not going to be a problem 18:24:28 so to be safe we should feed all videos into vocto before we play them... 18:24:35 maybe rather than try converting videos we should talk to vocto upstream instead and have the bug fixed? I'm starting to feel more & more that I'm doing things for no particularly good reason 18:24:41 that too, I guess 18:25:01 if there's a way to automatically test whether vocto would crash, I can run that script on the output of SReview, too 18:25:10 (automatically) 18:25:44 I still think it's a good idea to re-encode 18:25:48 sorry that I'm a bit behind on what's going on, but what does that mean, would users upload their videos through some frontend to sreview and then after some processing it gets streamed from there during DebConf? 18:26:01 we'll figure out why vocto crashes on that profile for sure 18:26:02 well, I'm not going to throw away the work I just did :) 18:26:08 wouter: automatic tests seems like a good idea 18:26:21 highvoltage: something along those lines, yes. I've modified SReview a bit so people can use it to upload recordings 18:26:41 highvoltage: we then run it through ffmpeg using SReview's infrastructure for that, so that all our files are encoded the same way 18:27:07 SReview just dumps the re-encoded files in a directory somewhere, we stream files in that directory to vocto during the conference 18:28:49 which brings us to... 18:28:55 #topic video stack - playback 18:29:10 playing files works from voctoweb (just tried it :) 18:29:10 nice transition :) 18:29:30 obviously if the encoding issues can be solved, but we already talked about that 18:29:39 yes 18:29:52 I don't know if we need to say a lot more about this 18:29:53 tumbleweed: ^ 18:30:10 yeah, it seems to be working fine 18:30:17 obviously some big missing features, like seeking 18:30:35 #info basic playback in voctoweb works 18:30:41 and no real organisation of the files, yet 18:30:59 but that could be as simple as just copying the right files for the day into the directory at the start of the day 18:31:13 well, this is related to the point about voctoweb above, so I guess we can talk about that later 18:31:13 if something happens in the middle of the video and we need to restart it in the middle, it's going to be hard without seeking :s 18:31:36 pollo: yes 18:31:54 we can always use the grabber input + vlc as a hotfix 18:32:01 but that machine would need access to the files 18:32:05 pollo: does that work with audio? 18:32:15 ah 18:32:16 also, that makes thing very complicated for volunteers 18:32:23 it could :) 18:32:30 anyway, another good transition: 18:32:36 #topic video stack - grabber 18:32:43 you can grab the default pulseaudio with x11grab, but we aren't doing that atm 18:33:04 we now have a grabber1.online.debconf.org machine 18:33:10 accessible via vnc 18:33:23 jitsi does grab audio from the video stream (and in fact mutes the microphone) if you switch to playing a youtube video 18:33:28 so I think it should not be an issue 18:33:42 we'll have seeking, don't worry 18:34:01 the grabber is available as a source in voctoweb 18:34:16 so it can be used to show basically anything 18:34:36 but as I said, currently it doesn't grab sound 18:34:39 should I add that? 18:34:47 sorry, what exactly is the grabber supposed to do? 18:34:52 I think I'm missing something 18:34:59 wouter: it's for questions on etherpad / gobby 18:35:12 it's a voctomix source 18:35:30 oh, we're not using jitsi for that anymore then? Or am I missing something? 18:35:40 wouter: it can potentially also used for things like a 'now/next' page (if someone makes one), or a note that the talk is delayed/cancelled 18:35:46 oh, that bit 18:35:49 okay, right 18:36:04 wouter: we don't plan to have attendees log to jitsi to ask questions no 18:36:30 wouter: having a design for such a webpage would also be nice :) 18:36:33 questions will go through pad / gobby and the talkmeister will as them 18:36:40 ivodd: heh :) 18:36:53 okay, clear. Thanks. 18:37:12 I think that's everything for the video stack 18:37:26 #topic streaming setup 18:37:34 no news here 18:37:40 it's ticking along happily 18:37:40 I don't think olasd is around, and everything was already working fine last week, so I suggest we move on 18:37:49 #topic Advice/training for directors (/talkmeisters) 18:38:04 as long as we don't have a working setup, I don't think we can do much here as well 18:38:18 yeah, things might change too much still 18:38:28 #topic Advice/training for presenters 18:38:29 although we could start an outline and general overview kind of thing? 18:38:54 that was finished already, I thought? 18:39:01 advice was done 18:39:20 highvoltage did say he wanted to add stuff ? 18:40:11 pollo: yeah it's just a few stuff I'll merge back to videoteam docs no big deal 18:40:17 ok, moving on 18:40:27 #topic Blocked salsa issues 18:40:36 I think all of them have been discussed already, so moving on 18:40:43 #topic Actions from last meeting 18:40:45 still have my gridengine MR 18:40:53 but that is not my highest priority 18:41:22 (as in, working on it, but prioritizing SReview) 18:42:12 wouter: anything we need to discuss about that right now? 18:42:20 three action items fro olasd (who's not here), and two for tumbleweed 18:42:28 ivodd: nope, just wanted to mention it so it ends up in the notes 18:42:34 wouter: ok, good :) 18:42:51 the action items for the people who are around are all discussed already, I'll copy the others to the agenda for next time 18:42:59 #topic any other business 18:42:59 yeah, was about to say the same :) 18:43:03 anyone? 18:43:21 do we have a schedule for when we want things to work? 18:43:32 not that I know of, but good idea 18:43:43 pollo: I was going to ask under 'test runs', but we can talk about it now 18:43:53 at least it would be good to have a semi-working setup 18:44:00 so that means feeding jibri into vocto 18:44:26 feels like we should be able to test things soonish 18:44:30 if we have that, we have all the sources in vocto, and can try a pre-run 18:45:04 does anyone have a suggestion for a timeline? 18:45:35 "be ready two days before debconf starts?" 18:45:36 what do we mean exactly by "test run" 18:46:09 (which means work our way back from then, figure out how much time everything needs (broadly) and set deadlines based on that) 18:46:30 pollo: ideally, it would be doing a test with volunteers to see if everything works 18:46:50 volunteers not being use core members who hacked on the system 18:46:52 pollo: I think we have people (preferably people who did not write the code in question) go through the motions of uploading, playing, etc stuff? 18:46:56 s/use/us/ 18:47:05 pollo: yes 18:47:18 so a pre-test run would be a test by us (who know details of the setup) 18:47:23 maybe we can ask urbec to do these things? 18:47:28 then we'll need 1 week to make a call for volunteers? 18:47:50 I think if we ask, we'll get some people quickly, but the issue is that we aren't ready for that yet 18:47:54 actually, no, she might know too much for that to be a useful test 18:48:16 is that true though? 18:48:23 voctoweb is mostly there 18:48:27 SReview is mostly there 18:48:35 we don't have a jitsi feed into vocto yet 18:48:45 we're probably close 18:48:49 is that a lot of work to finish? 18:49:00 I think if we ask volunteers now, that gives us a bit of a deadline, which might help 18:49:01 if paddatrapper can land the jibri MR, we can pre-test this weekend 18:49:13 and have volunteers test next week 18:49:21 right, that's what I was thinking 18:49:32 I think that's a good idea too 18:49:33 and then we can ask for volunteers once the jibri MR lands 18:49:55 I think we had a number of people who said the were happy to help testing a month ago 18:50:15 so I'll add the info later (under the right topic) 18:50:21 any other 'other business'? 18:50:26 good to know -- we should probably ask them then 18:50:40 maybe we should #agree something first? 18:50:58 #agree the jitsi bridge should be blue 18:51:02 sorry, been on the phone to family 18:51:19 #agreed ask for volunteers to test things after jibri MR lands (hopefully this weekend) 18:51:48 does that make sense? 18:51:55 +1 18:52:16 ok 18:52:25 #topic upcoming meetings - next irc meeting 18:52:37 same time same place next week? 18:52:38 next Thursday I guess 18:52:46 we seem to have a good rhythm going, would like to keep it 18:52:51 ok 18:53:01 #topic upcoming meetings - pre-test runs by video team 18:53:01 (even if thursdays are not ideal for me) 18:53:24 #agreed pre-test run by video team this weekend 18:53:27 (see above) 18:53:43 #topic upcoming meetings - test runs with volunteers 18:53:56 #info hopefully the weekend after next 18:54:11 also see above :) 18:54:16 right :) 18:54:19 anything else? 18:54:35 #endmeeting