>* Mimic absolute positioning via relative positioning. It's good as a short-term solution, but imho, we need something better
I'm not quite sure I understand which way you're describing: Do you mean having plasma code still do absolute co-ordinates and then have us turn it into relative positions like how Qt does popups? Or having client code do things in relative co-ordinates and locally convert it to absolute and using the existing plasmashell protocol. > The most ideal case is where Qt provides native support for the layer-shell protocol out of the box. Qt6 will have a new way of doing native functions on platform windows which looks very encouraging. Effectively it'll become QWaylandWindow: public QPlatformWindow, public QWaylandSpecificMagicWindowStuff Then Qt ships a header file of QPublicWaylandSpecificMagicWindowStuff which will contain all the getters with native code and a bunch of virtuals for setting things and one can downcast the platform window. Compared to the old nativeinterface approach it's /amazing/. I think that'll be a good entry point for all this. Then at a KDE level we can surface that as agnostic public API that does it's own local thing on X, and uses this Qt-platform-specific API on wayland. David
