>>> No.  sunlabel ought to create perfectly good big-endian labels even
>>> on little-endian machines, [...].  And the miniroot image is just
>>> an octet stream.

>>> Endianism would play a role only if you were trying to actually
>>> construct a filesystem on the little-endian machine [...]

>> I've succesfully used your sunlabel program, and a canned miniroot
>> to create a NetBSD/sparc boot disk from a NetBSD/i386 machine.

> So... if my labels are OK but all I need to do is newfs my disk, I
> can do this on a PC based *nix system?

No, that's what you _can't_ do.  If you already have a byte-for-byte
image of what you want to put on the disk (eg, the miniroot from the
install kit), it doesn't matter whether you use a 386 running DOS or a
Commodore 64 or a VM/CMS machine - bits is bits, and the Sun doesn't
care how they got there.

But in general, creating a filesystem with a non-native endianness is
not possible.  (I believe NeXT has done the necessary work, to support
sharing disks between black - ie, m68k - hardware and Intel hardware.
I don't know of any other vendor that has.)  Thus, running newfs on an
Intel box for use on a Sun _doesn't_ work.  The person who talked about
creating a SPARC boot disk was using a pre-generated miniroot image.

> Why oh why isn't newfs available in the tape boot media (asking
> innocently but completely ignorantly ;)

Not having been the one who decided that, I can't say for certain, but
I'd guess that the theory is that you're supposed to boot from tape,
partition the disk, and copy the miniroot from tape to disk.  Then you
boot from the miniroot and you have a stripped-down system that is
capable of newfsing and unpacking stuff.

					der Mouse

			       mouse@rodents.montreal.qc.ca
		     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B