15:44 <leon-p> ifreund: +1 for using u31 for widths. And I think we should also have an assert() in the code somewhere that enforces and documents the maximum window size. IIRC at the moment it's all a bit vague and not well defined and river will probably crash and burn when a window really does get that large.

15:46 <novakane> also +1 for u31, I started using it everywhere I can in my projects, really remove a lot of @intCast

15:46 <novakane> haven't come accros bugs with it for now

21:50 <leon-p> I think we can probably drop to something even lower than u31 if we wanted to. Even u16 would still hold larger numbers than possible on a 16k (!) screen, so realistically enough for the next five to ten years. Although I think there are fruits hanging a lot lower if we wanted to do some optimisations.

21:55 <ifreund> leon-p: safetly casting from i32 to u16 is more complex than simply checking for 0

21:56 <ifreund> and I don't think using u16s would make any noticible difference from an optimization standpoint

21:56 <ifreund> s/for 0/for >=0/

22:21 <kennylevinsen> 16k is also "only" 4x 4k screens side by side