Originally posted by avis
View Post
To be correct we have not split that far.
https://en.wikipedia.org/wiki/List_o...ervers#Wayland this is tip of iceberg.
Your development healthy wayland compositors.
You have your rust based.
https://github.com/Smithay/smithay So Cosmic is healthy because more than 1 party use it backend.
KDE is heathly because there are other compositors using the QT wayland backend so workload is shared.
Sway is healthy again because many parties are using wlroots backend so sharing the load.
Weston is heathly because every party who writes a desktop shell for weston shares the development of weston..
Enlightenment wayland development not healthy. Like who other than Enlightenment uses EPL
Mutter of gnome.... Avis who uses the libmutter library. for wayland other than Mutter.... So its not healthy.
Avis in wayland compositors if you are looking at healthy development speed. A compositor has to be Qt, Wlroots or smithay if you want to do your compositor as a single application or Weston desktop shell .so
Yes that 4 way code base split.
Qt is for those trained in C++ and C++ skills are what you need to maintain it.
Wlroots is C and you need C skills to maintain it.
Smithay is rust and you need rust skills to maintain it.
Weston is restrictive due to be reference compositor and you need C skills to maintain it..
Remember x.org code base is basically pure C. In reality the majority C++ and Rust programmers don't want anything todo with pure C.
Avis something you did not consider is a single code base can only target the developers to maintain it who like the language it coded in. KDE kwin is a Qt compositor so work on KDE and other Qt based compositors comes out of a pool of C++ developers. Cosmic with Smithay/rust the developers work on this come out of the pool of rust developers. Wlroots comes out of the pool of C developers that line up with the possible pool of developers for X11 bare metal server.
Having rust and C++ developers working on Wayland compositors because they have compositors in language they like using this increases the developers working on the Weston core protocols.
Yes Wlroots and weston do compete on developer resources to a point.
The forking equals more developers to review the Wayland core protocols.
Avis it would be a very different matter if the forking was like all C or all C++.... Because this would not be accessing different pools of developers.
One of the problems why X11 protocol development died is that forks of X11 server died.
X.org server what competition did it have avis to compete against. Think gcc before LLVM. Gcc massively improved in features after LLVM appeared once it had solid competition that could show that that X bad performance was in fact a issue with gcc because llvm could do it better.
Avis lack of competition is never a good thing. You are right there can be too much competition the art is getting the correct balance. Part of it is how can you have the biggest possible pool of developers work with core protocols so they can be designed as well as possible..
Comment