If anything, we can expect more people making Wayland compositors as hobby projects, if Waylands claims about a simpler codebase are to be believed.
They are not. Wayland compositors have to do a lot more of the same thing in every compositor than window managers ever had to. So many in fact that their whole central design idea has to be corrected for by everyone using wlroots to implement those common parts to get anywhere anyway which means wayland compositors in other languages without wlroots bindings are less likely.
have to do a lot more of the same thing in every compositor than window managers ever had to
Yes, but is that not entirely expected? As far as I understand, compositors are complete implementations of Wayland’s display server specification, whereas window managers are just a helper program that, well, manages windows, while Xorg does the heavy lifting required to fully implement the X Window System protocol. So the only real difference that I see is that, in the X world, the “common parts” are managed by a separate process (Xorg), whereas in the Wayland world, they are managed by a separate library (wlroots). So a hobbyist developer trying to make a window manager in some obscure language would need to figure out how to communicate with Xorg in that language, whereas a developer trying to make a compositor in some obscure language would need to write wlroots bindings for that language. Maybe I am just ignorant, but those seem like comparable efforts to me.
And lastly, in the X world, the only (widespread) implementation of the X Window System protocol is Xorg, but, in the Wayland world, there are compositors that use wlroots, and those that don’t. So wouldn’t that alone indicate more fragmentation / diversity? Sure, there are more X window managers than Wayland compositors out there, but X11 has also existed for longer. In short, I don’t see how the Wayland system is more adverse to diversity of implementations than X
They are not. Wayland compositors have to do a lot more of the same thing in every compositor than window managers ever had to. So many in fact that their whole central design idea has to be corrected for by everyone using wlroots to implement those common parts to get anywhere anyway which means wayland compositors in other languages without wlroots bindings are less likely.
Yes, but is that not entirely expected? As far as I understand, compositors are complete implementations of Wayland’s display server specification, whereas window managers are just a helper program that, well, manages windows, while Xorg does the heavy lifting required to fully implement the X Window System protocol. So the only real difference that I see is that, in the X world, the “common parts” are managed by a separate process (Xorg), whereas in the Wayland world, they are managed by a separate library (wlroots). So a hobbyist developer trying to make a window manager in some obscure language would need to figure out how to communicate with Xorg in that language, whereas a developer trying to make a compositor in some obscure language would need to write wlroots bindings for that language. Maybe I am just ignorant, but those seem like comparable efforts to me.
And lastly, in the X world, the only (widespread) implementation of the X Window System protocol is Xorg, but, in the Wayland world, there are compositors that use wlroots, and those that don’t. So wouldn’t that alone indicate more fragmentation / diversity? Sure, there are more X window managers than Wayland compositors out there, but X11 has also existed for longer. In short, I don’t see how the Wayland system is more adverse to diversity of implementations than X
I see wlroots as the bad workaround for the bad design decision to not have a single implementation in Wayland.