Xfce's Wayland Roadmap Updated

[creative]

avis

I think I may have tried LXQT once a long long time ago, it sounds familiar it may have been something else. I pretty much used a lot of older style stuff like Blackbox, Wmaker, and Fluxbox before I came across XFCE which was the center of a distribution I tried. I always thought Gnome and Kde were very slow even from the very beginning of my GNU Linux journey.

[Daktyl198]

X was designed for when applications asked the display server to draw 2d widgets for them, print for them etc, as it was considered that the user at the time would remote into a more powerful computer and stream a session.

This is not how things ultimately turned out, we use 3D APIs to render widgets into textures and then composite them onto the display. Every other modern operating system does this (Mac, iOS, Android etc). Windows is a weird one but simplified it does this.

Wayland is built for this model, there's nothing to "be outdated" because it's effectively a wrapper on how GPUs these days are built to work (passing around buffers/surfaces). Where as X had all the legacy baggage (drawing, fonts, 2d apis, printing etc) and was built for a different hardware model.

This is how compositing in Wayland is "passive", where applications are sending directly to the compositor/display server (which in some cases can use hardware planes to get no-copy outputs onto the display), where in X the compositor has to fetch the contents of each window from the display server, composite it, then send it back to the display server.

This is some crazy blind faith that technology will stay the same from now on. Computer graphics are literally changing right now with each generation, moving away from standard rasterization and toward specialized tensor cores. Who's to say that in another 10 years time, the graphics pipeline looks anything remotely like it does right now? Are we going to put layers and layers of shims on top of the Wayland API to make it keep working, or are we going to write a new API from the ground up again?

[mxan]

Wayland’s simple protocol, and the focus on moving functionality into the toolkits and other client-side libraries instead of doing so much in the display server itself, actually makes it easier to port to another windowing system if another one comes around in the future.

That said, I’m not sure if Wayland will actually be replaced any time soon. It’s core protocol was deliberately designed to be so tiny it’ll essentially be evergreen - the extensions change, but the core doesn’t.