Over time, a lot of useful stuff was broken and supported less and less, then reimplemented in bad way over compositors, instead of having composting Xserver like Xsgi.
Security extensions or even basic security tools (private input grab) got waylaid because toolkits didn't implement them (try running GTK3 on hardware without 32bit color GL or on system that defaults to lower bit depth than 32/24).
Generally, after letting it get more and more broken, the suggestion for big rewrite arrived. Said suggestion often results in something that resembles "devs first graphic stack", with a lot of basic, important features for daily life relegated to "someone will make an extension for it later" or "use D-bus to implement it elsewhere, and who cares about compatibility".
A bit like IRIS GL never getting open sourced, because it wouldn't be of much use without matching hardware exposed over /dev/GL interface.
Now we've reached the point where there are almost two parallel stacks - linux's graphics infrastructure; and Nvidia's. The developers don't have any patience for the one holdout against the new system. It is unfortunate for the people buying Nvidia cards because they are now starting to be labeled as 'using the legacy systems'. Presumably there is some reason why they have no choice in their purchases, but the cost of maintaining systems to work with closed source drivers appears to be too high. My views here aren't exactly current - I've stayed a long way away from Nvidia since ~2009 - but they are probably sill somewhat on target.
It is likely that a better designed alternative to Wayland will crop up. It is much easier to compete with than X - there are no drivers that only support Wayland.
I would have slightly preferred AMD, but I wanted an i9 in a laptop that had good Linux support, and so nVidia was all I could get (XPS15).
nVidia needs to choose to support Wayland, not Redhat.