[01:58] <RAOF> <grayson-g[m]> "I can't seem to sign into the..." <- Not sure why you can't log in, but I've updated that on the wiki page.
[05:37] <Saviq> Three issues I've found with Miriway+MATE on RC3 so far:
[05:37] <Saviq> - can't scroll all the way down, as if it's trying to scroll _past_ the bottom, and gets rejected
[05:37] <Saviq> - scrolling is jerky, compared to GNOME
[05:37] <Saviq> - notification appeared randomly in the middle of the screen (but that may be MATE's problem)
[05:46] <Saviq> - one of the symptoms of jerkiness seems to be that when I change scrolling direction, sometimes the first step will be in the original direction
[05:47] <Saviq> Sorry I've focused on scrolling, but that's something we've been talking a lot about recently, and is quite jarring UX wise.
[05:48] <Saviq> I can see those issues in Firefox, not in Fractal (Qt), that may be relevant
[06:55] <RAOF> Isn't Fractal GTK?
[07:09] <Saviq> You're right. Not sure why I thought Qt.
[07:11] <Saviq> Hah. That's special ;)
[07:11] <Saviq> $ xeyes
[07:11] <Saviq> Error: Can't open display: wayland-0
[07:16] <RAOF> Heh.
[07:17] <alan_g[m]> <Saviq> "I can see those issues in..." <- DISPLAY=wayland-0 is a MATE "feature"
[07:17] <Saviq> Yeah I grok it now, didn't yesterday when you said it :)
[07:18] <Saviq> Ever got thunderbird to run wayland? `MOZ_ENABLE_WAYLAND=1` isn't doing the trick :/
[07:18]  * RAOF should spin up a Miriway session.
[07:19] <alan_g[m]> Saviq: There is a post on discourse
[07:29] <Saviq> Can't seem to get Xwayland going, either. Is `enable-x11=true` correct in the config file?
[07:30] <alan_g[m]> It shouldn't be needed - the `miriway` script should do that for you
[07:31] <Saviq> And if it doesn't? I can't see Xwayland running, would it be spun on demand? Which DISPLAY, then?
[07:31] <alan_g[m]> DISPLAY should allocate normally. try :0
[07:32] <alan_g[m]> Gotta run errands before work BBIAB
[07:40] <Saviq> Yeah tried that, no dice
[07:41] <Saviq> - can't drag tabs in Firefox
[07:56] <alan_g[m]> `ls -lh /tmp/.X*`?
[07:58] <alan_g[m]> <alan_g[m]> "There is a post on discourse" <- https://discourse.ubuntu.com/t/using-firefox-wayland-on-mir-servers/19704
[07:59] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths closed [pull request #2560](https://github.com/MirServer/mir/pull/2560): Add a libmiral-bin metapackage
[08:04] <alan_g[m]> ...and we're *still* waiting on riscv64 in [ppa:mir-team/rc](https://launchpad.net/~mir-team/+archive/ubuntu/rc/+packages)
[08:12] <Saviq> > `ls -lh /tmp/.X*`?
[08:12] <Saviq> There's a lot, like up to X8...
[08:12] <Saviq> And yeah, :8 works
[08:13] <alan_g[m]> Have you crashed Xorg a lot since booting?
[08:15] <Saviq> No, I've restarted the shell a lot
[08:15] <alan_g[m]> But... that ought to clean up after itself
[08:19] <Saviq> Yup, it didn't. https://paste.ubuntu.com/p/5dzRZqm7CQ/
[08:19]  * Saviq will try !MATE components, maybe will have better luck
[08:20] <Saviq> FWIW b
[08:20] <Saviq> y restarting I meant killing miriway
[08:24] <alan_g[m]> Ah, not a normal exit by Ctrl-Alt-BkSp?
[08:24] <Saviq> That doesn't do anything, neither does "Logout" from the top menu
[08:25] <alan_g[m]> Try a long hold
[08:25] <Saviq> My config, for reference: https://paste.ubuntu.com/p/RnTCSx4QrZ/
[08:25] <Saviq> alan_g[m]: That worked
[08:26] <alan_g[m]> If you have any apps open, you need a long hold to force close
[08:27] <Saviq> Ah.
[08:32] -GitHub[m]:#mir-server- **[MirServer/mir]** bors[bot] closed [issue #2568](https://github.com/MirServer/mir/issues/2568): mir_unit_tests crashes in initialization on Arch
[08:32] -GitHub[m]:#mir-server- **[MirServer/mir]** bors[bot] merged [pull request #2569](https://github.com/MirServer/mir/pull/2569): Only call global_mock->XInitThreads() if global mock exists
[08:59] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths opened [pull request #2574](https://github.com/MirServer/mir/pull/2574): Fix include guard
[08:59] -GitHub[m]:#mir-server-  
[08:59] -GitHub[m]:#mir-server- > Oops!
[09:01] <alan_g[m]> Saviq ^ quick review please! (Is low risk, so I want this for ~rc4)
[09:42] -GitHub[m]:#mir-server- **[MirServer/mir]** bors[bot] merged [pull request #2574](https://github.com/MirServer/mir/pull/2574): Fix include guard
[10:06] <Saviq> Hmm I can't get focus in a thunderbird (xwayland) child (? - event creation) window
[10:06] <Saviq> It works with mail composition, so it looks like the calendar event window has some other properties? It's missing a title bar
[10:11] <alan_g[m]> There are still some Xwayland support issues in Mir (but worth raising one if it isn't already there). But isn't thunderbird GTK? And use Wayland?
[10:22] <Saviq> I couldn't get the snap to go Wayland (yet)
[10:24] <alan_g[m]> Oh, I'm using the deb. Grep the snap for `DISABLE_WAYLAND` as it is a popular way of causing pain
[10:59] <Saviq> Yeah it has that indeed
[11:04] <Saviq> Apparently for a reason…
[11:04] <Saviq> $ thunderbird
[11:04] <Saviq> Error: no DISPLAY environment variable specified
[11:04] <Saviq> Must've been built without wayland support, will try and rebuild
[11:05] <alan_g[m]> A shame, many snaps I've experimented with are "fixed" by unsquashfs and removing it
[11:05] <Saviq> And this (build) comes straight from Mozilla
[11:06] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths opened [pull request #2575](https://github.com/MirServer/mir/pull/2575): Rename miral::CommandLineOption to ConfigurationOption
[11:06] -GitHub[m]:#mir-server-  
[11:06] -GitHub[m]:#mir-server- > Candidate for cherry-picking to 2.9 as we've already broken ABI
[11:10] <Saviq> > A shame, many snaps I've experimented with are "fixed" by unsquashfs and removing it
[11:10] <Saviq> It did, my terminal environment must've messed something else up - it works fine now - in Miriway, too
[11:13] <alan_g[m]> 😂
[11:14] <Saviq> to15lip83ja
[11:14] <Saviq> And there goes the password...
[11:15] <alan_g[m]> No special characters? Be ashamed!
[11:17] <Saviq> We need a screen locker in Miriway :P
[11:17] <alan_g[m]> PRs welcome!
[11:29] <Saviq> alan_g did you see the scrolling issues I mentioned? those would be candidates for fixing for 2.9 IMO
[11:30] <Saviq> (if they're regressions, that is)
[11:30] <alan_g[m]> I saw you mentioned them. Was it only Firefox/X11 though?
[11:31] <Saviq> No X11
[11:31] <Saviq> Both Firefox and Thunderbird on Wayland
[11:32] <alan_g[m]> Will try to reproduce. (In a short while, back to drafting an article and want to finish the section)
[11:54] <Saviq> Looks like it only affects web views (as opposed to folder/email lists) inside Thunderbird, which would suggest there's two distinct APIs, and one of them is affected
[11:54] <Saviq> Likely something to do with smooth scrolling or so
[11:55] <alan_g[m]> Well, from what I've seen of GTK, it doesn't support wl_seat > 5, so there should only be "legacy" scrolling
[12:04] <Saviq> The views certainly behave different, on GNOME as well - lists scroll by abruptly jumping around, web views smooth the movement out
[13:08] <alan_g[m]> Does "on GNOME as well" imply it is a client-side problem, not Mir?
[13:09] <Saviq> I meant that it behaves differently on GNOME, which suggests it's not the same scrolling API? Or at least a different handling of it
[13:09] <Saviq> It behaves fine on GNOME, wrong on Mir
[13:12] <alan_g[m]> We *should* be "doing the same thing". Mir advertises wl_seat 8, Mutter 5, but GTK binds to 5 in both cases. I need to reproduce and grab some WAYLAND_DEBUG logs to see what differs
[13:16] <alan_g[m]> So, I open firefox, use the scroll wheel and see what?
[13:18] <Saviq> When you scroll, you can't reach either top or bottom
[13:18] <Saviq> When you change scrolling direction, sometimes the first "step" after direction change will be opposite
[13:19] <alan_g[m]> I do see that. So far, so good.
[13:20] <alan_g[m]> It almost looks as though the "paint" is for the previous scroll event
[13:27] <Saviq> But you can click on the right things, so it isn't, really
[13:29] <Saviq> I'd say the accumulation is going wrong? One step behind?
[13:29] <alan_g[m]> OK, so we're not sending the `axis_discrete()` messages needed for smooth scrolling
[13:31] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths opened [issue #2576](https://github.com/MirServer/mir/issues/2576): We don't send `axis_discrete` to ws_seat-v5 clients (like firefox)
[13:32] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths edited [issue #2576](https://github.com/MirServer/mir/issues/2576): We don't send `axis_discrete` to ws_seat-v5 clients (like firefox)
[14:25] <alan_g[m]> Weird, I saw the problem with the PPA build, but with a local build I don't...
[14:27] <alan_g[m]> But the accumulator code looks wrong. I think we need to use `std::div` to get sane results with signed values
[14:30] <alan_g[m]> ...or maybe not. Seems that was fixed (towards zero) in C++11
[15:34] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths closed [issue #2576](https://github.com/MirServer/mir/issues/2576): We don't send `axis_discrete` to ws_seat-v5 clients (like firefox)
[15:37] <alan_g[m]> OK, after reading what the logs actually say, the princple difference between us and Mutter is the order we send events in:
[15:38] <alan_g[m]> Mutter: source, discrete, axis, frame
[15:38] <alan_g[m]> Mir: discrete, axis, source, frame
[15:51] <Saviq> Does the protocol define the order? Or is Mozilla assuming something?
[15:52] <grayson-g[m]> Saviq: I don't recall the protocol specifying order.
[15:54] <grayson-g[m]> "The order of wl_pointer.axis_discrete and wl_pointer.axis_source is not guaranteed."
[15:59] <alan_g[m]> It shouldn't matter. And changing our order doesn't change anything
[16:29] <alan_g[m]> OK, there's a factor here I don't follow. The same scenario with the same build of the same code that was failing consistently is now working consistently
[16:36] <Saviq> I can reliably see it with my build (against ~rc4)
[16:42] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths opened [issue #2577](https://github.com/MirServer/mir/issues/2577): [2.9] Weird scrolling in, for example Firefox... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/c7c3eeede985dbbfc377d4609072244e0992ac9b)
[16:43] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths edited [issue #2577](https://github.com/MirServer/mir/issues/2577): [2.9] Weird scrolling in, for example Firefox
[16:44] <alan_g[m]> I've seen it (and not seen it) with both PPA and local builds.
[16:48] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths edited [issue #2577](https://github.com/MirServer/mir/issues/2577): [2.9] Weird scrolling in, for example Firefox
[16:49] -GitHub[m]:#mir-server- **[MirServer/mir]** AlanGriffiths edited [issue #2577](https://github.com/MirServer/mir/issues/2577): [2.9] Weird scrolling in, for example Firefox
[17:02] <alan_g[m]> OK, that's it for my day. o/
[17:09] -GitHub[m]:#mir-server- **[MirServer/mir]** bors[bot] merged [pull request #2575](https://github.com/MirServer/mir/pull/2575): Rename miral::CommandLineOption to ConfigurationOption
[19:40] <Saviq> sophie if you have an idea, would be good if you have a look ^^
[19:41] <sophie-w> Saviq: What are you referring to, the FF scrolling thing?
[19:44] <Saviq> sophie yes
[19:45] <sophie-w> Yep, that's my after lunch project (been finishing up a fix for the deadlock I found the other day)
[19:46] <sophie-w> (sorry if the FF thing was urgent, I though y'all had logged off)
[20:02] -GitHub[m]:#mir-server- **[MirServer/mir]** wmww opened [pull request #2578](https://github.com/MirServer/mir/pull/2578): ObserverMultiplexer: allow simultaneous observations/unregisters... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/29385a82c6dae9f6f4451ef63eca6f2723e03680)
[20:02] -GitHub[m]:#mir-server- **[MirServer/mir]** wmww requested a review from RAOF for [pull request #2578](https://github.com/MirServer/mir/pull/2578): ObserverMultiplexer: allow simultaneous observations/unregisters
[20:06] -GitHub[m]:#mir-server- **[MirServer/mir]** wmww closed [issue #1880](https://github.com/MirServer/mir/issues/1880): ObserverMultiplexer: deadlocks if observer removed from callback when executor runs immediately
[20:17] <Saviq> No worries, not urgent, just didn't want it to miss your day :)