> After enough time some best practices do solidify into standards, but it wasn’t overnight in X either, and first and foremost it wasn’t made with that in mind.

Part of what drives me nuts about Wayland is they could have learned these lessons from X's history and actually had it in mind. Instead, they threw it all out without learning a thing.

Or chances are, the more features they would have demanded by spec from the start, the smaller the number of people actually interested in implementing it would have gotten, and it would turn into a single implementation at most, or just hype at the worst.

They absolutely chose to do the protocol with enhanceable protocols, something greatly missing from Wayland, so I don’t think your criticism is fair here.

This is a result of them learning lessons from X history. The lesson from X is that mouse configuration absolutely does not belong in the display protocol and should not require futzing around with editing a root-owned xorg.conf. Input configuration is an entirely separate concern.

And? Just because the linux ecosystem is fragmented doesn't mean that new software has to be, too.

The fact that developing a new compositor with wayland basically required wlroots to be even feasable for a one-man project is concerning enough. The amount of stuff you have to handle is insane compared to writing your own x-wm.

That's not a good comparison because it would be even harder to write your own X server. And yet X is still extremely fragmented.

And developing a new compositor also does not require wlroots. Originally there was libweston, wlroots is just another option.

It is a good comparison because you don't need to write your own X server to implement your own window manager, which is the part that most people care about.