Wayland protocol objects versioning & XDG Shell initial size #2254
Conversation
| // We don't care about obscure pre version 4 compositors | ||
| // and just damage the whole surface instead of | ||
| // translating from buffer coordinates to surface coordinates | ||
| self.wl_surface.borrow().damage(0, 0, i32::MAX, i32::MAX); |
There was a problem hiding this comment.
druid actually uses the damaged region in order to optimize its own drawing, so I think if you damage the whole surface here then you'll end up with a partially rendered window. Which is probably no big deal if all real compositors support this. But I think it at least deserves a warning...
There was a problem hiding this comment.
Would self.invalidate() do the job? The same way we do it for force == true a few lines abowe.
There was a problem hiding this comment.
Yep, I guess that should take care of telling druid that it needs to draw everything. I'm a little confused about that invalidate() though, because shouldn't we also be calling self.wl_surface.borrow().damage(...) in that branch?
There was a problem hiding this comment.
I think it should be called, as far as I can tell what happens when no damage is sent is not specified in the protocol. So compositor may just do nothing in that case.
|
Thanks for the PR! It would be great if the wayland backend received consistent love :) The CI is complaining that you need to run |
…der#2254) * wayland: Fix WlSeat versioning * wayland: Fix XdgShell versioning * wayland: Fix XdgToplevel initial size * wayland: Fix WlCompositor versioning * wayland: Fix WlOutput versioning * Address review comments
This PR introduces support for running on a wide variety of versions of core protocols.
Druid can now run on both Weston and Mutter.
While at it, I also fixed the initial size of top levels, they are no longer 1px x 1px, they now use the size provided by the user.