[08:37] <Saviq> More sticks.
[08:37] <Saviq> https://forum.snapcraft.io/t/does-snap-refresh-trigger-the-disconnect-connect-hooks/11997
[08:38] <Saviq> -> No.
[08:40] <Saviq> Our multi-glvnd-driver plan is not working very well at the moment :'-(
[08:52] <Saviq> Or any ancillary setup, for that matter — as you can't know when the "leaf" snap was refreshed to refresh whatever you expose to the consumer snap
[10:05] <alan_g[m]> So... we would need a daemon sitting there watching for changes?
[10:10] <Saviq> I don't think that's worth it. Until/unless SnapD lets us do it with hooks, I'd just bite the bullet and maintain Mesa and Nvidia+Mesa snaps. If someone is pressed for space, a Nvidia-only snap could also make sense (and, btw, when you're pressed for space the ancillary approach would be even worse since you have to copy).
[10:16] <alan_g[m]> Yup, just wondering if there were any options
[10:17] <Saviq> TBH the best solution would be if content was "passed through", then no copying or hooks would be needed. Something to ask SnapD if they would ever consider
[10:18] <alan_g[m]> Yes, but AIUI mounts can't escape the namespace
[10:25] <Saviq> Well, yeah, SnapD would need to add explicit support for that
[10:36] <Saviq> Oh and the other part is, it would be appropriate for the Mesa and Nvidia snaps to slot both the glvnd-driver-core22 interface and graphics-core22 - this way if you needed only one, you'd use it directly. But then we're back into SnapD getting in a bind situation, because when you _do_ use glvnd, you've multiple graphics-core22 providers.
[12:22] -GitHub[m]:#mir-server- **[MirServer/ubuntu-frame]** bors[bot] edited [pull request #129](https://github.com/MirServer/ubuntu-frame/pull/129): [Merged by Bors] - Allow configuration to assign applications to outputs
[12:22] -GitHub[m]:#mir-server- **[MirServer/ubuntu-frame]** bors[bot] closed [pull request #129](https://github.com/MirServer/ubuntu-frame/pull/129): [Merged by Bors] - Allow configuration to assign applications to outputs
[15:48] -GitHub[m]:#mir-server- **[MirServer/mir]** bors[bot] closed [issue #2857](https://github.com/MirServer/mir/issues/2857): Popup positioning does not seem to respect constraint adjustment
[15:49] -GitHub[m]:#mir-server- **[MirServer/mir]** bors[bot] merged [pull request #2858](https://github.com/MirServer/mir/pull/2858): Implement XDG poup constraint adjustment support
[16:29] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths opened [issue #2859](https://github.com/MirServer/mir/issues/2859): 23.04 slow startup of GTK applications... (full message at <https://libera.ems.host/_matrix/media/v3/download/libera.chat/c1c94f0c92c155c6b443d48d0b0b00d4aa84c677>)
[17:45] <alan_g[m]> Saviq I'm still tweaking, but this might give some you some relief after your fight with snapd: https://github.com/Miriway/Miriway/pull/22
[18:58] <Saviq> OK a win at least. Got a mesa-core22 that works on Classic with Nvidia. Weird to be shipping the eglexternalplatform in there, but…
[18:59] <Saviq> And o/
[20:31] <Saviq> RAOF (he/they) if you could please try things out: `mesa-core22/edge/pr2` together with `graphics-test-tools/edge/pr1`. As far I can tell, everything works as well as it can on classic (note there's `.vainfo` and `.vdpauinfo` now and `.glmark2-es2` now).