bolphen

> I think we both agreed `KeyCode` should have a layout-independent behaviour (because of the `WASD` example). Therefore, the Linux behaviour sounds like a bug to me by definition. Actually...

> I think we should put extra care here. Native scancodes and virtual keycodes are definitely going to make the API worse. I believe we should stick with some `enum`...

I think this is a gnome-specific thing: see https://discourse.gnome.org/t/5243 TL;DR gnome uses `WM_CLASS` and the `.desktop` entry to determine the icon. So miniquad needs to set a correct `WM_CLASS`, which...

I think the `keys_down` and `mouse_down` should be cleared on the [`window_minimized`](https://github.com/not-fl3/macroquad/blob/master/src/lib.rs#L794-L797) event, which is the closest we have to a `focus_lost`. ```rust fn window_minimized_event(&mut self) { get_context().keys_down.clear(); get_context().mouse_down.clear(); #[cfg(target_os...

These are already being cleared every frame in `Context::end_frame`

In practice this is ok since globally there is only one instance of `X11Clipboard` guarded behind a `Mutex`. Although I agree that I don't like the static mut either --...