Re:I'm tired of this anti-Wayland horseshit
by cycloptic on Tuesday February 02, 2021 @02:17PM
(#26005272)
See the protocol extension I linked, there are absolutely exceptions and escapes that can fall back on the X11 way of doing things, and could even work with such X11 client isolation, but somebody would have to take the time to actually design and implement them and figure out how to get them to work with the security model used by the Wayland implementations. That is really the problem, it's only an implementation detail if you choose to do it by hacking the X server or you choose to make a new D-Bus API for it, or whatever. (i.e. there is no real reason XWayland needs extra extensions, it could call a D-Bus API if necessary, and that may even be desirable because the D-Bus API would be forward-compatible with newer Wayland applications)
This is inimical to the design principles of Wayland, whose security policy is to isolate all clients -- no exceptions, no escape.
The 1984 Mac was largely a novelty, but it was one that demonstrated possibilities. There are non-novelty applications that are precluded by Wayland. Something like EXWM could not be done. You cannot write a standalone virtual-desktop switcher, it must be integrated into the compositor. Et weary cetera. Wayland is very opinionated about the kinds of programs you can write for it, and if you want to go beyond that you must traverse the step function into "write your own compositor".