| -GitHub[m]:#mir-server- **[MirServer/wlcs]** RAOF opened [pull request #250](https://github.com/MirServer/wlcs/pull/250): CI: Drive LXD directly, avoiding spread | 07:08 | |
| RAOF | Saviq: what do you think of dropping the use of spread for WLCS and just driving the tests directly? | 08:39 |
|---|---|---|
| RAOF | I know why Ubuntu 22.10+ are failing, and could add another hack into spread to work around it, but it'd just be another hack, and we don't actually use most of the features of spread… | 08:41 |
| Saviq | Oi! | 08:52 |
| Saviq | RAOF (he/they) one reason why I like spread (at least for Mir) is that we (now, through LXD) can trivially reproduce the builds locally. What would the hack be? There is _some_ movement in spread upstream so we could try and move things there | 08:54 |
| alan_g[m] | Does it gain much over using LXC to set up a test environment locally? | 09:00 |
| Saviq | With the breadth of configs, IMO it does, since any config is just a `spread lxd:…` away. Maintaining that set locally is possible, but by the time you actually need it, you're likely to need to update enough for it to be easier to start from scratch | 09:02 |
| Saviq | While we could build a tool of our own around it, would it be better? | 09:03 |
| alan_g[m] | If we were doing it more often, I would care more. But it is an infrequent inconvenience | 09:21 |
| Saviq | "it"? | 09:21 |
| alan_g[m] | testing some non-native environment locally | 09:22 |
| Saviq | Right, but we'd have to maintain it for CI | 09:22 |
| Saviq | And I currently doubt it would be less maintenance than our spread solution | 09:23 |
| alan_g[m] | That is a more convincing argument than "can trivially reproduce the builds locally" | 09:31 |
| -GitHub[m]:#mir-server- **[MirServer/mir]** Saviq assigned AlanGriffiths to [issue #2695](https://github.com/MirServer/mir/issues/2695): Miriway died upon disconnecting external screen | 09:51 | |
| -GitHub[m]:#mir-server- **[MirServer/wlcs]** Saviq opened [issue #251](https://github.com/MirServer/wlcs/issues/251): Support for private tests... (full message at <https://libera.ems.host/_matrix/media/v3/download/libera.chat/c3673b374e6277f15ddf5e222e979d21c9ff71d3>) | 10:53 | |
| RAOF | @Saviq A fix (for now) for the spread failure-to-set-up-22.10+ images would be to delete `/etc/ssh/sshd_config.d/60-cloudimg-settings.conf` before restarting sshd in setup. | 11:27 |
| Saviq | alan_g on display reconfiguration, if through a wayland extension, would it be possible to switch layouts without creating a surface? That's a limitation of the sleep inhibitor that we would want to do away with some time. | 11:42 |
| Saviq | On that note, should sleep timeout maybe become part of the display layout? Sounds like something we may expose per display? | 11:42 |
| alan_g[m] | There's no requirement for a surface, just a connection on which globals can be published. But we'd obviously want to use `wl_output` | 11:45 |
| Saviq | FWIW I believe there are two distinct features here - clients asking for a display configuration (that we don't want on Frame, by default), and switching between predefined display layouts. The latter is what we should focus on. | 11:48 |
| alan_g[m] | The sleep inhibitor protocol chooses to use a surface so that inhibition can be associated with surface visibility. You could have a different protocol that does things differently | 12:15 |
| Saviq | Sure, just wanted to clarify what we're aiming for. Working with `wl_output` to select layouts sounds wrong to me when a layout describes more than one output | 12:16 |
| alan_g[m] | Thinking back to how mirclient managed display configurations we may want to allow a surface to be associated with a display configuration | 12:21 |
| Saviq | Sure, for the general case, but Frame should not respect those requests by default. We have a static display config defined by the operator and we should never stray from it | 12:23 |
| alan_g[m] | We have enough control over publishing extensions for that | 13:03 |
| -GitHub[m]:#mir-server- **[MirServer/wlcs]** Saviq requested a review from RAOF for [pull request #252](https://github.com/MirServer/wlcs/pull/252): spread: use spread-mir-ci | 15:50 | |
| -GitHub[m]:#mir-server- **[MirServer/wlcs]** Saviq opened [pull request #252](https://github.com/MirServer/wlcs/pull/252): spread: use spread-mir-ci... (full message at <https://libera.ems.host/_matrix/media/v3/download/libera.chat/49bc0201f495d5d326f898fc5f12c2cce244cd01>) | 15:50 | |
| alan_g[m] | > fix unqualified use of `std::move` | 16:17 |
| alan_g[m] | Why would that be needed? | 16:17 |
| Saviq | https://github.com/MirServer/wlcs/actions/runs/3516098418/jobs/5892209519#step:5:1164 | 16:18 |
| alan_g[m] | "-Wunqualified-std-cast-call` wtf?! | 16:29 |
| Saviq | Good night all o/ | 17:54 |
| alan_g[m] | o/ | 18:13 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!