07:50 Company: why is it wl_surface::set_buffer_transform and not wl_buffer::set_transform?
10:15 kennylevinsen: Company: scale is also set on the surface rather than buffer. I suppose you could have had such state on all buffers in your swapchain, but I don't think that would have felt nicer...
10:44 Company: kennylevinsen: I'm trying to understand the pros and cons of doing that so I can judge where best to put that property in GTK
10:47 Company: scale is kinda different because there's fractional scales and scale is often done together with offset (like in wp_viewport with the source/dest rects)
10:47 Company: while the transform ultimately describes in which direction to read the buffer memory
10:49 Company: and there's zwp_linux_buffer_params_v1::y_invert
10:49 Company: which overlaps with the buffer transform - unless I'm missing something
10:50 Company: actually, for flipped buffers, should one use zwp_linux_buffer_params_v1::y_invert or wl_surface::set_buffer_transform - is that worth specifying somewhere?
10:53 zamundaaa[m]: Company: always use the surface one
10:54 zamundaaa[m]: At least wlroots, Weston and KWin don't support the buffer flag anymore
10:56 Company: don't support that flag or don't support any flag?
10:56 mclasen: if the flags were a bad idea, maybe they should be dropped from the protocol
10:57 zamundaaa[m]: Company: KWin at least doesn't support any of the flags. Import will always fail if any is set
10:57 Company: the protocol is finally stable as of a few months ago!
10:58 Company: but it's probably worth deprecating them or at least adding a big warning to the docs
10:58 zamundaaa[m]: Yeah it does seem like an oversight that needs fixing
12:04 any1: Is there any way to change LED state of a keyboard other than pressing a lock key on the keyboard?
12:06 vyivel: any1: libinput_device_led_update()?
12:11 any1: vyivel: I meant if it was possible through a wayland protocol. I haven't found anything, but I just wanted to be sure.
12:12 vyivel: ah, no, don't think there's anything like taht
12:14 any1: Well, that's good. :)
20:26 vaxry: hey guys, I wanted to make my compositor's event loop reading async, so instead of wl_display_run I'd poll the wl_event_loop fd and do wl_display_flush_clients and wl_event_loop_dispatch on a POLLIN, but that for some introduces stutters and stuff like slow app open times, any ideas what's wrong? https://github.com/hyprwm/Hyprland/commit/f2a848cbcc41f29fb62ee67aef95136ae1a650da
20:27 vaxry: my lock-stuff logic should be sound as I use almost the same in a bunch of other places, plus, moving to polling 2 fds instead and still being single-threaded didn't change anything