Victoria Brekenfeld

@Supreeeme I fixed the issue for cosmic here: https://github.com/pop-os/cosmic-comp/pull/1824 You might want to adopt the XSETTINGS xwayland-satellite is setting similarly. The key was adjusting `Gdk/UnscaledDPI`.

> [@Drakulix](https://github.com/Drakulix) Could you give the reference implementation in KDE? It seems that only change in GDK unscaled dpi doesn't work and MATLAB seems to only respect XFT dpi. I...

> > I am very certain that at least the configuration in cosmic-comp works > > You mean that, if I don't misunderstand it, `matlab` just scales up with the...

I would prefer to fix CI first and then merge some already approved PRs.

You can configure that in `~/.config/cosmic/com.system76.CosmicComp/v1/xkb_config`

> > This is because Smithay, the base component of cosmic-comp, doesn't strictly follow the protocol. > > The cursor position set by clients with `text_input_v3::set_cursor_rectangle` requests should be double-buffered...

Possibly relevant: https://github.com/Smithay/smithay/pull/1388

> @Drakulix the rejection that I experienced happens here: > > https://github.com/Smithay/smithay/blob/4fbcc9252088825dbf1e3f900c99ae04df5053d6/src/wayland/selection/data_device/device.rs#L82 > . > Since smithay did not know of an ongoing pointer-grab it doesn't start the dnd. That...

So the issue is, that we don't start a `ClickGrab` inside popup for the last mouse down event, because we already have a `PopupGrab`, thus the serial isn't matching. What...

Relaxing the check is also not correct, given that the protocol expects an implicit click grab explicitly and not anything else. We could potentially get away by internally updating the...