10:56 <skuzzymiglet> hey all, my screen sometimes freezes on river, usually due to inactivity

10:57 <skuzzymiglet> not sure how long before it freezes, it can be 2-5 seconds after inactivity

11:03 <ifreund> sounds like this thing: https://github.com/ifreund/river/issues/153

11:04 <ifreund> I kinda expect it to just disappear once I implement damage tracking

11:05 <skuzzymiglet> what's the progress on that?

11:08 <ifreund> I'll probably get to it shortly after river-layout is finally mereged

11:09 <ifreund> which is currently blocked on me deciding how best to handle configuration of "default" output options

11:09 <skuzzymiglet> I've developed a muscle memory of ctrl-alt-2 ctrl-alt-1-ing to fix it now :)

11:09 <ifreund> heh, I admire your persistence :D

11:10 <ifreund> there's also the fact that I personally have never hit this so I'm not particularly motivated to work on it yet...

11:13 <skuzzymiglet> I could take a shot if you want

11:14 <skuzzymiglet> don't know the codebase and I've only dabbled with Zig though

11:16 <ifreund> I mean if you want to go for it I wont stop you. Damage tracking isn't exactly trivial though, it it were I'd have already done it

11:18 <leon-p> very simplistic start-stop damage tracking actually is trivial, but I am not sure it is enough of a performance improvement to be worth it

11:18 <leon-p> (source: some experiments with zig-wlroots)

11:23 <ifreund> might still fix #153 by using the more tested wlroots code paths for damage tracking

11:41 <leon-p> IIRC here is what to do if anyone does wants. It's not much more than creating wlr_output_damage for every wlr_output and using its frame events instead of the outputs one. Then in the frame handler do wlr_output_damage.attachRender, which will tell you if you need to render a frame. If not, wlr_output.rollback(), if yes, render like we do now. Now you need to add wlr_output_damage.addWhole() everywhere

11:41 <leon-p> where the sreen gets dirty.

11:41 <leon-p> s/does wants/wants to/g

11:41 <leon-p> s/sreen/screen/g

11:42 <leon-p> AFAIK wlroots takes care of dirty-fying the screen for output mode changes, but I am not sure

12:09 <ifreund> the "add wlr_output_damage.addWhole() everywhere" part is the hard part