18:54 <leon-p> ifreund: I remember you once saying you wanted to write a native zig wayland implementation. Have you started that already? I thought about forking zig-wayland, but if there already is a project, contributing to it would make more sense.

19:08 <emersion> leon-p: are you aware of the gotchas with third-party wayland impls?

19:09 <emersion> mostly about interoperation with other libraries which use libwayland, such as OpenGL, Vulkan, wlroots

19:10 <ifreund> leon-p: nope, and I'd recommend starting from scratch instead of forking zig-wayland

19:10 <leon-p> emersion: yes, I am

19:10 <ifreund> the whole point of a from scratch impl would be to have a better API

19:10 <leon-p> ifreund: I was thinking of recycling the scanner though

19:10 <emersion> okay, just making sure :)

19:10 <leon-p> emersion: thanks anyway :)

19:10 <ifreund> ah yeah the xml parsing code is certainly reusable

19:11 <ifreund> emersion: if I understand correctly, it's theoretically possible to get hardware acceleration without libwayland if you reimplement whatever you need directly using linux-dmabuf

19:12 <emersion> yes, correct

19:12 <ifreund> I don't have a strong sense of how infeasible that is though

19:13 <emersion> you can probably translate wlroots' wayland backend + gles2 renderer + swapchain + allocator + …

19:13 <emersion> well, not translate because you don't need the whole wlroots API and architecture, but pick the pieces

19:13 <ifreund> yeah, it's all the same machinery

19:14 <ifreund> leon-p: prior zig wayland without libwayland art: https://github.com/tadeokondrak/zwl

19:14 <ifreund> also https://github.com/malcolmstill/foxwhale

19:15 <ifreund> neither are very completed but may be useful as a source of API inspiration

19:15 <leon-p> Ah, thanks

19:16 <leon-p> yeah the API will be the most challanging aspect. My plan was to get it to a point where I can safely iterate the API for a few months until I like it

19:17 <ifreund> I think it would be nice to have even if hardware acceleration is too involved to reasonably implemented

19:18 <ifreund> software rendering is totally adequate for many clients, and foot proves that they can be perfectly fast

20:21 <euclio> Seems like the river/tags module of Waybar isn't working anymore. Is this a known issue? Would it be on the Waybar or the river side?

20:23 <msiedlaczekelshi> euclio: iirc you need to compile waybar from master currently; the fix is probably still not available in your packaged version

20:24 <euclio> Ah, that was it. Thanks!