14:05:20 <cschalle> #startmeeting Workstation WG
14:05:20 <zodbot> Meeting started Mon Feb 11 14:05:20 2019 UTC.
14:05:20 <zodbot> This meeting is logged and archived in a public location.
14:05:20 <zodbot> The chair is cschalle. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:05:20 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:05:20 <zodbot> The meeting name has been set to 'workstation_wg'
14:05:32 <cschalle> #meetingname workstation
14:05:32 <zodbot> The meeting name has been set to 'workstation'
14:05:40 <cschalle> #topic Roll call
14:06:23 <cschalle> #chair mclasen otaylor stickster aday
14:06:23 <zodbot> Current chairs: aday cschalle mclasen otaylor stickster
14:06:52 <juhp> .hello petersen
14:06:52 <zodbot> juhp: petersen 'Jens Petersen' <petersen@redhat.com>
14:07:03 <cschalle> #chair ryanlerch
14:07:03 <zodbot> Current chairs: aday cschalle mclasen otaylor ryanlerch stickster
14:07:47 <cschalle> #chair juhp
14:07:47 <zodbot> Current chairs: aday cschalle juhp mclasen otaylor ryanlerch stickster
14:08:06 <bcotton> .hello2
14:08:07 <zodbot> bcotton: bcotton 'Ben Cotton' <bcotton@redhat.com>
14:08:21 <cschalle> ok, I am not sure who is chairing today (could be me)
14:08:39 <juhp> I thought it was stickster, but dunno if he is around :)
14:08:54 <otaylor> The schedule shows stickster
14:09:13 <juhp> nod
14:09:20 <mclasen> .hello mclasen
14:09:21 <zodbot> mclasen: mclasen 'Matthias Clasen' <mclasen@redhat.com>
14:10:11 <cschalle> ok, so assuming stickster is not around, we got 1 item on the agenda today and that is about gamemode being installed by default
14:10:20 <cschalle> #info https://pagure.io/fedora-workstation/issue/88
14:10:37 <cschalle> #link https://pagure.io/fedora-workstation/issue/88
14:10:44 <otaylor> Also, figuring out what we're doing with the meeting time :-)
14:11:38 <cschalle> ok, well I also think we might be missing so many members we can't actually hold any votes
14:12:06 <cschalle> but I guess we can discuss and see what the general feel is for those two issues
14:12:35 <cschalle> so to start with the gamemode discussion. Anyone here who would object to the proposal for including it by default?
14:13:34 <otaylor> Don't really object, I do feel we're writing some weird blank check for doing whatever tweaks seem to help framerate without any review with how it fits into kernel/systemd/etc. policy
14:14:14 <otaylor> But pragmatically seems OK for now
14:14:23 <cschalle> otaylor, well we do of course still hold the right to refuse an update or fork it
14:14:28 <mclasen> I still don't like the architecture of that thing, but...
14:14:39 <otaylor> cschalle: But practically speaking, we're *not* going to be reviewing changes one by one
14:14:57 <otaylor> unless it's caught red handed doing something bad, it's just going to follow upstream
14:16:19 <otaylor> I certainly don't have time to spend to create systemd-gamemoded ... so that limits my objection here
14:16:33 <cschalle> true, but since 'everyone' is shipping it these days, there should be a fair amount of eyeballs on it
14:17:05 <mclasen> it doesn't fit into flatpak, really
14:17:25 <otaylor> mclasen: you mean, the architecture?
14:17:33 <mclasen> yes
14:17:36 <otaylor> mclasen: But a 'gamemode portal' could back end to it?
14:17:53 <mclasen> its a library that does random syscalls, which we likely want to block in the sandbox
14:18:11 <cschalle> mclasen, wouldn't it be on the host and not in the sandbox?
14:18:24 <otaylor> mclasen: I thought the random syscalls were in the sandbox? Or is there a client-side piece too?
14:18:29 <otaylor> errr, in the daemon
14:18:31 <juhp> This only affects gamers?
14:18:36 <mclasen> its a library, no ?
14:18:40 <mclasen> libgamemode
14:18:54 <mclasen> if its on the host, its not going to help flatpaks at all
14:19:10 <mclasen> anyway, I don't like it, but thats not a vote
14:19:40 <cschalle> mclasen, well it would if we had a portal like otaylor suggested right?
14:20:11 <mclasen> what sort of portal is that "random tweaks to make this game run faster" ?
14:20:51 <otaylor> mclasen: the portal is "I'm running a game now, do appropriate stuff"
14:21:13 <otaylor> mclasen: what the appropriate stuff is, isn't part of the portal API
14:21:37 <mclasen> can you really ask the user: "BLABLA wants to run as a game. [Allow][Deny]" ?
14:21:43 <mclasen> is that meaningful ?
14:21:54 <cschalle> mclasen, the games automatically turn it on
14:22:01 <otaylor> mclasen: well, the first question would be, is that necessary - if the downside is only that it kills your battery, it probably doesn't need to be interactive
14:22:40 <otaylor> (since apps can kill your battery already by just running at 400% cpu)
14:23:10 <mclasen> so yes, having a dbus api behind it would be the way to make this function in sandboxes
14:23:22 <mclasen> still not convinced its a great api
14:24:37 <otaylor> mclasen: talking to prarit, it was pretty clear that you can't get the same thing by just tweaking scheduling algorithms, so the alternative is probably trying to detect "game is running" in mutter, but that's going to be fiddly and not much fun
14:24:45 <cschalle> mclasen, well look at it like this, we are not likely to get into investing in creating great gaming APIs, and the major companies around linux gaming like Ferral and Valve are behind this API and want to use it, so unless we have spare cycles to engage I feel we should assume they have experience with optimizing your system for gaming
14:25:21 <mclasen> they are not creating great gaming apis either :(
14:25:40 <mclasen> assuming is never a good strategy
14:25:40 <cschalle> mclasen, but if you are not going to write games yourself, is it really your problem? :)
14:26:31 <mclasen> you are making it mine, by asking me to agree
14:26:37 <juhp> lol
14:27:08 <cschalle> mclasen, I am sure we ship hundreds of libraries whose APIs could be subjectively better :)
14:27:34 <otaylor> I'm not as optimistic as cschalle, but our investigation earlier didn't show anything really crazy going on ... and given the lack of resources as Workstation WG to create the ultimate api here...
14:27:54 <mclasen> yeah
14:28:16 <mclasen> anyway, its still not going to be useful for flatpak games unless somebody finds the resources to make that portal
14:28:26 <juhp> true
14:29:03 <juhp> Still adding the library seems fairly harmless, no?
14:29:12 <otaylor> Perhaps, given lack of quorum we should move on? (do we just have mclasen, cschalle, juhp, and myself, or is everybody else being quiet?)
14:29:19 <cschalle> anyway, lets shelve this until next time as we don't have voting majority here today
14:29:22 <mclasen> yes
14:29:39 <cschalle> ok, so meeting time. Any suggestions for the change?
14:30:35 <otaylor> I think we just need to use whenisgood or whatever that thing is, and see what works - we're fairly constrained with europe + juhp so either some other morning works, or not.
14:30:57 <cschalle> ok, anyone volunteer to set up a whenisgood and try to find a time?
14:31:19 * mclasen can do an hour earlier
14:31:38 <otaylor> I can do that
14:31:44 <cschalle> otaylor, ty
14:31:48 <juhp> Thanks
14:31:49 <cschalle> ok, I think thats it for today then
14:32:03 <cschalle> #endmeeting