The input stack on Linux: An end-to-end architecture overview
venam.net93 points by venamresm__ 6 hours ago
93 points by venamresm__ 6 hours ago
How is how inputs gets routed to the right window out of scope for an article that wants to describe input end to end? The fact that input events get carefully routed to the right thing is both important and a potential source of bugs if not implemented correctly.
> How is how inputs gets routed to the right window
It is covered for both X11 and Wayland. I just don't get into the particular decisions and details of how each WM/DE picks what they deem the current focused window, since it varies widely and it's more part of window management than of input management (I've written a WM and it's a bit messy). An article on WM/Compositor development would be more appropriate for that, I have a few already on my blog.
It's not covered.
>Obviously, each have their own internal representation and ways of managing the information they get from libinput, XKB, and others, but this is outside the scope of this article
From what I can tell this is the explaination. It says that they manage input. If this is good enough why not simplify what the kernel does by saying Linux has its own way of handling inputs from keyboard and that is out of scope. It just seems arbitrary what handling of input is and isn't in scope.
That’s the domain of a window manager. It’s touched on in the article, but going into detail about the edge cases would require detours into user space window manager choices and even preference settings.
I'm not even talking about edge cases. It doesn't even give a high level. It just treats wayland and x as big black boxes and does not explain what happens inside of them.
Very impressive, nice work!