(In reply to comment #4)
> I notice that if /tmp is unwritable (e.g. due to a full disk) the X server will
> not fully come up and startkde is never executed, so ksmserver will not have a
> chance to tell anything to the user.
>
> So I suggest that kdm adopts some of the sanity checks from ksmserver/main.cpp
>
wrong. startkde IS executed, I just tested this scenario.
(In reply to comment #4)
> I notice that if /tmp is unwritable (e.g. due to a full disk) the X server will
> not fully come up and startkde is never executed, so ksmserver will not have a
> chance to tell anything to the user.
>
> So I suggest that kdm adopts some of the sanity checks from ksmserver/main.cpp
>
wrong. startkde IS executed, I just tested this scenario.