07:03 <tsujp> ifreund im trying to build river on void from your draft PR but getting `ninja: error loading build.ninja no such file or directory'

07:04 <tsujp> i think the build-style might be incorrect, its set to use cmake instead of zig-build?

07:06 <tsujp> whoops dw thats the zig template

07:06 <tsujp> okay so this is a zig build problem..

07:49 <dnkl> ifreund: leon-p: I've updated the foot PR (for the CSD/SSD bug). It now assumes there will be an xdg-surface configure event following the decoration configure event, and applies the CSD/SSD change at that point.

07:49 <dnkl> this appears to fix all CSD/SSD issues on Sway

07:51 <dnkl> re. the number of sub-surfaces: there are 4 invisible ones, for drag-to-resize. The titlebar (which is actually title-less) is the fifth. Then I got lazy, and implemented the minimize/maximize/close buttons using sub-surfaces. So that's 8 in total.

07:51 <dnkl> we don't _need_ sub-surfaces for the buttons, but it makes it so much easier to implement the button hover effects, and click-detection.

10:13 <ifreund> tsujp: not much I can do to help without seeing error messages...

10:14 <ifreund> you may be interested in this PR adding a zig-build xbps build style https://github.com/void-linux/void-packages/pull/29288

10:57 <ifreund> dnkl: Just confirmd that your foot PR fixes the xdg_surface geometry size change when csd state is updated on river

11:06 <dnkl> ifreund: great, thanks for testing!

12:31 <ifreund> wlroots 0.14.0 is released, time to break CI again

12:31 <ifreund> Or maybe I'll stop being lazy and start building wlroots from source

12:34 <emersion> please do :^)

12:36 <leon-p> ifreund: since xdg-activation is now in wlroots, do we want to have it in river 0.1.0 ?

12:36 <leon-p> if yes, its behaviour should be consistent with the xdg foreign toplevel activation request

12:37 <ifreund> leon-p: I'm not going to block river 0.1.0 on it, I just want to fix the bugs and cut a release

12:37 <ifreund> we can do an 0.1.1 fairly soon adding more features if it makes sense

12:37 <leon-p> makes sense

12:37 <ifreund> and now that wlroots 0.14.0 has released, I think we should do river 0.1.0 asap

12:38 <leon-p> agreed.

12:49 <leon-p> I think I can solve #337 and another bug related to it today

12:52 <ifreund> nice!

12:52 <ifreund> I'll be taking a look at #329 after I bump the wlroots version

12:53 <leon-p> #337 is actually somewhat nasty. Cursor code totally ignores that the output or tag focus can change, even during an operation.

12:57 <ifreund> so what, we need some Cursor.resetState() function that we call when changing outupt/tag focus which handles things?

12:57 <leon-p> exactly

12:58 <leon-p> although I'll have to redesign some things since most functions in Cursor.zig take events.

13:31 <novakane> ifreund: have you decided for unstable/staging btw?

13:32 <ifreund> novakane: I'm not going to break waybar

13:33 <ifreund> next time we do a breaking change to the current protocols using the old naming scheme we can switch to the new one

13:34 <ifreund> got wlroots building from source in the CI, wasn't too bad really :D

13:34 <ifreund> when then next zig version releases we can switch to using the static zig tarballs in the CI too

13:35 <ifreund> and then CI should never break again

13:38 <novakane> makes sense

13:38 <novakane> > and then CI should never break again

13:38 <novakane> sure... :P

13:39 <ifreund> someone should add musl CI runs

13:39 <ifreund> I don't know if github actions can do freebsd, but that would be nice too

13:40 <novakane> ifreund: https://github.com/marketplace/actions/freebsd-vm

13:40 <novakane> looks like so

13:46 <leon-p> ifreund: we could also use sourcehutes CI if githubs does not work.

13:53 <novakane> leon-p: he'll need to pay for that no?

13:55 <leon-p> I have a paid account, we could use that

13:57 <novakane> oh allright then let the boss decide :P

13:59 <leon-p> Was just an idea in case GH does not support FreeBSD CI, since I know sourcehut does support it.

14:02 <leon-p> submitted a draft PR. Not sure if it is the right approach, but it does fix the cursor image problem.

14:03 <leon-p> the problem is leaving .down state, because for notifying the client we'd need an event.

14:03 <leon-p> But I am tempted to just not notify the client. That would match the behaviour of basically every desktop and operating system I ever used.

14:11 <ifreund> Not notifying the client might be totally fine, I wonder what other compositors do

15:11 <leon-p> huh, I think when the layout changes, the update to the cursor state is false

15:12 <leon-p> it probably uses the wrong coordinates.

15:13 <leon-p> I have the feeling we might need to update the cursor state on every transaction

17:02 <skuzzymiglet1> ifreund: I think foot in the void repos needs updating from 1.7.2 to 2.4.1

17:05 <ifreund> skuzzymiglet1: are you talking about the revbumping foot for the fcft update? that shouldn't be needed. fcft 2.4 is ABI compatible with 2.3 afaik, and the soversion didn't change

17:05 <ifreund> (and it works for me)

17:06 <skuzzymiglet1> oh, confused foot and fcft release pages :)

17:11 <dnkl> skuzzymiglet1: ifreund: patience, I'll release 1.8 any day now :)

17:12 <skuzzymiglet1> :D