[07:34] G'morn [10:41] Saviq we do want to cherry-pick #2697 into release/2.10? [10:42] Yeah was planning to, along with anything else coming in. But I think the biggest question is https://github.com/MirServer/mir/pull/2645 - do we want to wait for that, or do we push with the release as is? [10:44] That optimisation can go in another release when it works [10:45] If you think it really urgent I could take a look at fixing it up, but I don't think it is needed [10:49] OK, I also pulled https://github.com/MirServer/mir/pull/2689 in [10:49] What about https://github.com/MirServer/mir/pull/2688 ? [10:50] It is unlikely to cause problems, and might fix some. I'd leave it, but wouldn't object [10:50] There's no clear issue associated with it, so I'd leave it be [10:51] deal [11:25] Well, I can confirm that using the same setup I use for Miriway, Frame only displays the wallpaper on one screen:... (full message at ) [11:25] So I will go and dig [11:25] But not something x11 platform specific, at least [11:32] But can't reproduce when nesting it on Miriway through Xwayland 🤦 [12:43] And now got it under Miriway, too. It must be a timing issue, does the internal background listen to displays coming and going? [13:09] alan_g I can, however, reproduce your input problems on my second laptop [13:09] I don't know whether to be pleased or displeased [13:12] Wonder if this is relevant `[ 415.844212] rfkill: input handler enabled` [13:14] It is printed whether it works or not, so unlikely [13:17] "And now got it under Miriway..." <- Here's another (probably timing related) issue: https://github.com/MirServer/mir/issues/1981 [13:25] "I don't know whether to be..." <- Adding this to the profile helps, after a bit of playing in the dark:... (full message at ) [13:25] But I don't get any more detail from apparmor about which it actually uses [13:27] So it is AppAmor [13:27] Actually just one asterisk is fine [13:28] And really `/sys/dev/char/* r,` is enough [13:29] So, that is just wierd [13:30] +many [13:31] The only device it actually opens here is drm, but it suggests libinput gets lost if it can't enumerate or something [13:40] So, why is this 22.10+hardware specific?! [13:42] May be some newer driver for Gen 12th hardware. But why wouldn't AppArmor report it. [13:44] Anyway, I guess we need to PR a change to the Wayland slot. [14:54] I doubt that will get accepted, though, `/sys/dev/char/*` is a bit broad… [14:55] True, but it is a curated slot [14:56] Well, actually not I suppose [14:59] I'll ask on the forum, maybe they'll help dig deeper [15:29] https://forum.snapcraft.io/t/keyboard-mouse-input-not-working-on-22-10-and-missing-apparmor-info/32157 [15:36] Is it also worth mentioning that input is supported by the `wayland` slot? (For Frame we also have the `hardware-observe` plug to reduce log spam) [15:39] Done. [15:55] Ah. So we only respect the `output_id` of a fullscreen window. But we spoof it to be maximized to respect exclusive zones. [15:56] Spoof it in Frame, I mean [15:57] Things seem to work for swaybg because the output_id gets carried through in WindowInfo [15:58] Now that you describe it, that does sound familiar [15:59] https://github.com/MirServer/ubuntu-frame/blob/058551a72bf5abedc87946d6f54f0fd202e7e799/src/frame_window_manager.cpp#L93-L95 [16:00] But https://github.com/MirServer/mir/blob/fbc60bee416149e164a25b23c8c5114339aecd07/src/miral/basic_window_manager.cpp#L1307-L1311 [16:12] Ok, more tomorrow. o/