[07:02] <Saviq> Mornin'
[07:41] <alan_g[m]> Good morning!!
[08:11] -GitHub[m]:#mir-server- **[MirServer/mir]** RAOF opened [issue #2570](https://github.com/MirServer/mir/issues/2570): Wayland platform uses EGL 1.5 entry points
[08:11] -GitHub[m]:#mir-server-  
[08:11] -GitHub[m]:#mir-server- > Some EGL implementations (still!) don't support EGL 1.5. We don't _really_ need anything beyond 1.4, but https://github.com/MirServer/mir/blob/eebf0ff47353b8d4c8f7636144cb4e3248be5afd/src/platforms/wayland/displayclient.cpp#L331 is an EGL 1.5 entrypoint that isn't provided by some EGL implementations.
[08:15] -GitHub[m]:#mir-server- **[MirServer/mir]** RAOF edited [issue #2570](https://github.com/MirServer/mir/issues/2570): Wayland platform uses EGL 1.5 entry points
[09:57] -GitHub[m]:#mir-server- **[MirServer/mir]** Saviq opened [issue #2571](https://github.com/MirServer/mir/issues/2571): Screen capture on early start... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/b508e6adbd63ff539a4ac06cb4d8a30013dab901)
[09:57] -GitHub[m]:#mir-server- **[MirServer/mir]** Saviq opened [issue #2572](https://github.com/MirServer/mir/issues/2572): Wheel-scrolling over remote assistance breaks input... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/556bee9f85677d9217050553ace0885dcd7f1e91)
[09:58] <Saviq> alan_g IMO we need to look at ^ for 2.9
[09:58] <Saviq> Mostly the latter, maybe not the former
[10:02] <alan_g[m]> There shouldn't be a race with ubuntu-frame-vnc, that probably needs some WAYLAND_DEBUG logs and/or the code around "main.c: 605" to see what fails
[10:04] -GitHub[m]:#mir-server- **[MirServer/mir]** bors[bot] merged [pull request #2566](https://github.com/MirServer/mir/pull/2566): Handle the new libinput events
[10:04] <alan_g[m]> "interface 'wl_pointer' has no event 9" that sounds like incorrect protocol extension versioning in Mir, I'll take a look
[10:04] -GitHub[m]:#mir-server- **[MirServer/mir]** bors[bot] closed [issue #2565](https://github.com/MirServer/mir/issues/2565): libinput warns of client bug when I scroll my wheel
[10:06] <alan_g[m]> And yes, the release should wait in investigating (and, hopefully, fixing) #2572
[10:07] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths assigned AlanGriffiths to [issue #2572](https://github.com/MirServer/mir/issues/2572): Wheel-scrolling over remote assistance breaks input
[11:03] <Saviq> alan_g @alan_g:matrix.org I wasn't able to reproduce without VNC, could it be that the events are delivered differently? They do go through wayvnc after all
[11:05] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths closed [issue #2572](https://github.com/MirServer/mir/issues/2572): Wheel-scrolling over remote assistance breaks input
[11:07] <alan_g[m]> Saviq you can reproduce by running a Mir desktop (I recommend Miriway). I couldn't get webkit to show a window on GNOME
[11:08] <alan_g[m]> Hmm, that negotiates "wl_seat, 5"
[11:08] <Saviq> Right, I was testing through Frame on GNOME, could well have been Xwayland in the mix
[11:09] <alan_g[m]> I don't think the X11 (or Wayland) platforms support the fine-grained pointer stuff at all
[11:21] <alan_g[m]> We're likely going to find dragons by allowing version 8. GNOME on 22.10 is only supporting version 5, so clients are unlikely to have been tested against the more recent revs
[11:58] <alan_g[m]> Saviq any progress on #2571 that would argue against just cherry-picking #2566?
[12:00] <Saviq> Having trouble reproducing now, but regardless don't think it's anything re #2566
[12:08] <alan_g[m]> I was more thinking of "more to cherry-pick", but I guess it is too soon
[12:28] <Saviq> alan_g looking at wpe code, here's where it binds to wl_seat:
[12:28] <Saviq> https://github.com/Igalia/cog/blob/d70401fc9ab4290dc2aaed5ff933d7eaf0cc746d/platform/wayland/cog-platform-wl.c#L770
[12:28] <Saviq> And it binds to version "2" explicitly, unless I'm missing something
[12:29] <Saviq> And that'd be https://github.com/Igalia/cog/pull/397
[12:30] <Saviq> So that changed relatively recently
[12:44] <Saviq> > Having trouble reproducing now, but regardless don't think it's anything re #2566
[12:44] <Saviq> Ah. Resizing the Frame window is part of the reproducer
[12:48] -GitHub[m]:#mir-server- **[MirServer/mir]** Saviq edited [issue #2571](https://github.com/MirServer/mir/issues/2571): Screen capture on early start
[12:58] <Saviq> And it looks like a wayvnc issue to me. It captures frames without clients being connected, and it's not about "early start", but rather resizing while connected
[13:03] -GitHub[m]:#mir-server- **[MirServer/mir]** Saviq closed [issue #2571](https://github.com/MirServer/mir/issues/2571): Screen capture on early start
[13:04] <Saviq> So looks like RC3 is good to go
[13:30] <alan_g[m]> Huh!? Miriway (on 22.10) has lost it's background with rc3
 "So looks like RC3 is good to go" <- Should be. I've done the basic checks. It is still worth letting everyone on the team "kick tires" first in case another wheel falls off.
[16:26] <sophie-w> Uh oh! found a deadlock. Initial signs point towards something to do with surface observers and/or screen capture. Investigating further...
[16:28] <sophie-w> I was investigating the wayvnc compositor resize bug at the time. This deadlock will distract me for a while, but it was looking like the issue may have been introduced in this commit from a few days ago: https://github.com/any1/wayvnc/commit/f3247ad9338d8abe19d6c9a56f2c45a94f9ba96c
[16:59] <grayson-g[m]> Miriway is supposed to build if I'm running 2.9.0 rc3 of `libmiral-dev`, right? it shows i have that version of libmiral but the build fails on `split_command()`
[17:00] <alan_g[m]> Do you have an older local instal?
[17:02] <alan_g[m]> Where is `pkg-config --cflags miral` getting the headers from?
[17:02] <grayson-g[m]> `-I/usr/local/include/miral -I/usr/local/include/mircommon -I/usr/local/include/mircore`
[17:04] <alan_g[m]> You've a local instal (in /usr/local) that overrides what you instal from archive
[17:04] <grayson-g[m]> alan_g[m]: Oop! I'll get that removed. Thank you!
[17:06] <alan_g[m]> OK, EOD for me!
[17:09] -GitHub[m]:#mir-server- **[MirServer/mir]** wmww opened [issue #2573](https://github.com/MirServer/mir/issues/2573): Deadlock while resizing Mir-on-X... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/198a7348b8ae108c5ffe814705f4ebf0f67dfe51)
[17:13] <Saviq> And me. Good night! (from Miriway, with dual displays how I need them now) o/
[19:17] <grayson-g[m]> I can't seem to sign into the wiki (I believe this has happened before but RAOF (he/they) could get in) but on https://wiki.ubuntu.com/Process/Merges/TestPlans/Mir under the `--console-provider=vt` note it mentions to set `XDG_RUNTIME_PATH` but I think the correct name is `XDG_RUNTIME_DIR`.
[21:44] <grayson-g[m]> I got Miriway built and poked around with RC3 throwing as many different programs as I could at it and it held up fine. Nothing fun to report.