On Mon, Sep 8, 2008 at 11:50 AM, Bernhard R. Link
<address@hidden> wrote:
* J.R. Mauro <address@hidden> [080908 15:08]:
> Now maybe I have a broken install or a broken understanding, but tmpwm just
> runs the new windowmanager and
> doesn't do any kind of recovery when the window manager exits to get RP back
> as your WM.
It actually does steps to get everything back, but there exists a race
condition in ratpoison that it sometimes tries this before the X server
has realized the other window manager already exited, so that reclaiming
control fails. (Most window managers need a little time after
disconnecting from the X server before finaly terminating, but when the
window manager is very fast in terminating or killed instead of properly
shutdown, it happens more often).
Interesting. Where is this code located? I'll take a look.
> As it stands, I kill the
> temporary WM and then run ratpoison -c "tmpwm /usr/bin/ratpoison". Is this
> the way we're supposed to do it or am
> I missing something here?
Instead of tmpwm newwm would be better, as it does not start ratpoison
as temporary window manager within ratpoison but replaces it with a new
copy. (If your other window manager also support this you can also
switch between the two managers using this way without tmpwm).
I thought this was hackish :)
My other WM is evil, which does very little for you, so I have to exec RP manually.
I've not properly analysed it (race conditions are an ugly thing), but
I think there can be different things that fail. Either ratpoison can
fail at all (most unlikely) then even ratpoison -c will no longer work.
if ratpoison -c still works, then only the key grabbing might have
failed, then even other commands instead of tmpwm or newwm can perhaps
give you back an working ratpoison.
I think this is the case. IIRC, the C-t hook worked, but my windows were not being managed
by RP.
Hochachtungsvoll,
Bernhard R. Link