07:25 <lordmzte> I've noticed that just recently my setup's screen configuration (scale, size, rotation) is lost after a suspend. Looking at the issue tracker, it looks like this is expected, but in that case, any ideas why these settings used to be persistent for me without any daemon?

07:26 <ifreund> lordmzte: wlroots behavior change around when wlr_output objects are created/destroyed

07:27 <lordmzte> Interesting. Thanks!

07:29 <ifreund> basically, all river users probably want to be running kanshi :D

07:29 <lordmzte> The two tools I've seen mentioned in the issue tracker (way-disaplys and kanshi) both seem to manage output arrangement. Is it possible to simply restore the old behavior, making wlr-randr changes persistent?

07:34 <ifreund> nope, river doesn't persist output configuration itself so something else has to

07:37 <lordmzte> And is there a tool that can do that? Those I mentioned all seem to "override" configuration done with wlr-randr

07:39 <vyivel> one could probably have a daemon that receives updates on output states and remember latest ones

07:40 <ifreund> I don't know of an existing thing that does that

07:45 <lordmzte> -_-

10:29 <lordmzte> Well alright, now that I've set it up kanshi seems pretty damn cool :P

14:32 <leon-p> I can reliably make rwm stutter when moving windows by pointer, simply by opening snayk (my snake clone) and resizing it once. Works in rivercompat as well, although only if the nested rwm isn't fullscreen

14:42 <ifreund> leon-p: interesting, does it have to be snayk?

14:42 <leon-p> currently trying to find out. Foot is fine, gtk stuff is fine

14:42 <leon-p> nothing interesting in wayland_debug sofar

14:50 <ifreund> leon-p: clang thinks it's found a bug in snayk by the way

14:50 <ifreund> https://bpa.st/raw/X6FA

14:50 <leon-p> yeah, snayk is from way before I discovered -fanalyzer, asan or anything like that :)

14:51 <leon-p> it's basically just a test client for me right now

14:52 <ifreund> defintely feels like a snayk bug to me at first touch

14:52 <ifreund> it's probably not committing after ack in some cases or something

14:52 <leon-p> probably, but I do wonder why it causes river to stutter. especially during move, where snayk shouldn't do anything

14:53 <ifreund> well, the transactions are timing out

14:54 <ifreund> perhaps river should start ignoring a client after consecutive timeouts until it acks again

14:54 <leon-p> I kinda expected it already does, so probably yes

14:54 <ifreund> but yeah, from the WAYLAND_DEBUG=1 snayk is doing exactly what I said, ack_configure with no commit

14:55 <leon-p> I should clean it up, maybe, eventually

14:55 <ifreund> also, river tries not to send a configure if nothing changed

14:56 <ifreund> however if the window manager proposes dimensions river always sends a configure currently

14:56 <ifreund> and at least the wms I've written so far are very liberal with their propose-dimenisions requests

14:58 <leon-p> river also currently sends wl_surface.enter/leave to all decoration surfaces when the window moves, even if the surfaces did not change output

14:58 <leon-p> btw, am I missing something or is there no otherway in zig-wayland to set the user data ptr other than attaching a listener?

14:59 <ifreund> perhaps river can not send a configure in response to propose_dimensions in some cases, but there are tricky races with timed out configures

14:59 <ifreund> leon-p: yeah, if there was a way to set it some other way it would break the type safety

14:59 <leon-p> fair

14:59 <leon-p> although typesafety goes away when you manually access it anyway

15:00 <ifreund> not sure what you mean

15:01 <ifreund> yeah, getUserData() returns an opaque pointer, but the callback itself has a type-safe signature at least

15:01 <ifreund> it's not perfect, but it's a decent middle ground

15:02 <ifreund> making every wayland object type generic over the user data type would suck

15:02 <leon-p> was just a bit confused that I have to attach a wl_surface listener despite not caring about the events to set the user data :)

15:02 <leon-p> but I get it

15:03 <ifreund> not sure about the wl_surface.enter/leave thing, river's probably holding wlr_scene wrong

15:04 <ifreund> really, I need to get the rwm branch merged so we can start using the issue tracker to track these things

15:05 <ifreund> or you know, feel free to just start using the issue tracker for this stuff with an rwm tag

15:07 <leon-p> probably a good idea, yeah

15:07 <leon-p> the reason I report most things on IRC is because the 1100 PR takes about 1 minute on my sad little laptop to load and get to a stage where I can write a comment

15:08 <leon-p> it's still 20 seconds on my office workstation, which is otherwise massively overpowered

15:08 <ifreund> yeah, I showed it to a codeberg person a few months ago and they said something like "that's a lot of database queries"

23:33 <Guest83> Hey all, just getting river setup. I have the default init setup and installed foot however I cannot launch foot