https://bugzilla.gnome.org/show_bug.cgi?id=762468
--- Comment #21 from Jonas Ådahl <[email protected]> ---
(In reply to Matthias Clasen from comment #20)
> what you see in the stacktraces is that we get a window state event which
> changes the fullscreen state, then terminal-window updates its toggle action
> to match, which in turn triggers the callback to call
> gtk_window_[un]fullscreen. Seems fine so far, since we are just calling the
> function for the state we already changed to, but something ought to notice
> this and fizzle out, probably on the mutter side.
What could mutter do? When it receives the make-fullscreen request, it
immediately responds with "ok, go fullscreen then". Then when it receives the
unmake-fullscreen request, it immediately responds with "ok, go unfullscreen
then".
I have a patch which "compresses" GDK_WINDOW_STATE events by dropping any
previous event to the same window while updating the "changed_mask". That seems
to fix the issue to me (though it triggers another bug, which I suspect is in
mutter, causing the window to not display any more, and then display with the
wrong size).
Another possible solution I intended to try is to only allow dispatching one
wayland event at once before dispatching the internal GDK event queue, but this
would have a lot larger impact on things.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
wayland-bugs mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/wayland-bugs
