> has anyone else noticed this "problem"? seems to me that the kernel or the
> xserver can't tell if there's a monitor attached like MacOS can (or if it
> can tell, it doesn't do anything about it.)

Monitor sensing isn't in there just yet. My intvid code will take care
of this on supported machines. It does a full monitor sense to detect
the type of monitor. If it comes up with no monitor (you're running
headless) the internal video driver doesn't bother attaching.

We could do this on NuBus cards too but it would require hardware support
for the cards. This may eventually happen as I have some crazy ideas
of porting my universal video driver interface to the NuBus driver.
We might be able to do the sensing through ROM calls on cards but
I don't think most early NuBus cards do the right calls in software
to perform monitor detection.