QtCS2021 - Wayland client BOF

Session Summary

Synchronization on topics concerning the Wayland client plugin with Qt, KDE and others.

Session Owners

• Eskil Abrahamsen Blomfeldt (eskil.abrahamsen-blomfeldt@qt.io)

Notes

Things to discuss:

- The upcoming "is occluded" protocol upstream (sounds amazing):

https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/99

- everyone agrees.

- The new Qt shell - what's the story?

Just for embedded. We (KDE) can ignore it.

- Our LayerShellQt library - creating API upstream and including XdgPositioner

- we need to configure properties before creating the shell

- we create the shell on show

- which doesn't give a hook

- either need to change qwaylandwindow (which is hard due to reasons)

-or set properties in advance

Agreed it would be good to have support for this in Qt.

- DataDevice V3 (can I poke for review)

Poke noted by Qt people, they'll follow up. Maybe I need to check too.

- Activation (poke? )

- embedded users

Poke noted.

Do we need it in compositors? Mixed results. Agreement that we can have a client-side implementation first and implement compositor-side when someone asks for it.

Generally lots of agreement that we need to start adding public API: https://bugreports.qt.io/browse/QTBUG-94729

> Qt Client side decoration paths; Can we shove them into a subsurface (or 4)? I think it would be cleaner

Everyone agrees. We need to be careful if we need to use place below as we might hit compositor bugs.

We should 4-tile though, it'll be more performant. Maybe we can make a POC and have Jan confirm it works and finish the details as he's the main user.

> Exposing fractional scaling? Good idea?

Yeah. Lets give it a go. How do we handle values that aren't multiples of buffer size? enums / common denominator? It will have issues, but no more than X11...lets give it a go and see.