Now that we have them, would it be feasible to use LLMs to go after the historical crud that X11 accumulated due to age?
I don’t like vibe coding, but using LLMs to dig into a huge legacy code base like X11 could be very useful.
X11's problems were rooted in the abstractions presented by the X11 core protocol and its extension mechanisms. The interface, not merely the implementation.
Wayland was correct in first focusing on replacing this interface. The problem is the effort stopped there and left the ecosystem to figure out the implementation part.
Broadly, the X Server has a bunch of capabilities which are irrelevant. The modern model is really Window <-> Compositor based, and the X Server protocol is just a pointless middle man in that exchange.
"compatibilities which are irrelevant" do exist, such as old drawing primitives but those are not really an issue. They can be maintained for backwards compatibility and eventually deprecated and removed, but they are not anything which holds back modern clients. The X server protocol is not anymore a "pointless middle man in that exchange" than the Wayland protocol is a "pointless middle man". A protocol is obviously needed so can not be pointless.