>>> Wouldn't it be possible to have /dev/sdXX be links to
>>> /dev/scsi/cXtXdXsX or somesuch ?
>> Sure, it would be possible, but that method is ... sort of ugly and
>> brings into question "Just who makes those symlinks?"

Right on both counts, though I'd replace "sort of" with "very".  In
fact, I'd go further and call it really gross.  One of the things I
hate most about our SGI machines is these long baroque disk names.

> Actually I guess the link could be done by the kernel itself (instead
> of internally mapping the /dev/sd0a to scsi-id 3, partition 1 at boot
> time, it could create a link on the file-system at the very same time
> and then forget about that mapping).

Um, the root filesystem is mounted read-only at that point, and may
never go read-write.  Besides, I (for one) really _really_ dislike
wiring knowledge of pathnames like "/dev/scsi" into the kernel.  The
only way I could excuse this is if /dev is wholly synthetic, a la
procfs or kernfs at present, and the kernel comes up with it mounted.
Which has plenty of problems of its own.

>>> Of course, dropping that hardcoded root drive (and swap drvies)
>>> would be good too.
>> Huh?  I'm not quite sure what you're saying here...
> I'm talking about having to reompile the kernel when you want to add
> a second swap drive,

Yeah, this is a bit ugly.  I'm sure whoever finally gets around to
fixing it will have a bunch of people grateful.

> or the (supposed) inability of the normal kernel to mount anything
> else than /dev/sd0a on /.

I think "config netbsd swap generic" (you actually want "root generic",
but they go together - something else that needs fixing) works now,
though I'm not sure.

> The root drive should be the same as the boot drive rather than being
> hardcoded in the kernel.

Well, it should be possible to configure it that way.  It should also
be possible to force / regardless of the boot device.  (Which way the
pre-built kernels should be done is an issue I don't want to comment
on; I really don't care, 'cause I always build my own kernel anyway.)

					der Mouse

			    mouse@collatz.mcrcim.mcgill.edu