On 9/4/06, Anselm R. Garbe <[EMAIL PROTECTED]> wrote:
Actually that is a suboptimal fix. I pushed a different solution
which uses the knowledge, that arrange() calls restack()
already. The described problem occured because 'sel' may
point to a different (the previously selected client) until
the focus() is called in the end, this results in a wrong Z
layer order when arrange() calls restack().
--- a/client.c Mon Sep 4 12:23:41 2006 +0200
+++ b/client.c Mon Sep 4 17:02:13 2006 +0200
@@ -254,6 +254,8 @@ manage(Window w, XWindowAttributes *wa)
clients = c;
settitle(c);
+ if(isvisible(c))
+ sel = c;
arrange(NULL);
XMapWindow(dpy, c->win);
XMapWindow(dpy, c->twin);
No, that's the way it was (I removed that part in my patch), and it
leads to another problem: besides focusing the new client, focus()
also makes sure that the previous client is cleanly unfocused (correct
buttongrab etc.). Changing sel from outside focus() leads to the
former selected client having incorrect buttongrabs. This could be
fixed by doing something like:
if(isvisible(c)) {
focus(NULL);
sel = c;
}
for instance, or by manually grabbing the right buttons for the old
focused client. Not sure what the right way is (calling focus on c
itself at that point is probably not an option, since that would
incorrectly map its twin afterwards).
Gr. Sander.
