At the risk of introducing a "dual-space" flame...

On Wed, 24 Apr 1996, der Mouse wrote:
# You still haven't addressed what to do with disks with port-specific
# compatability disklabels that don't _have_ types...like the sparc and
# sun3 ports, which support Sun-style disklabels, which don't have typed
# partitions.
# 
# Should the compatability code just return "undefined" for all its
# partitions?  That renders the boot disk (often the only disk) incapable
# of supporting these nice make-sure-I-don't-shoot-myself-in-the-foot
# things you've been touting....

Apparently, this isn't quite true.  Solaris supports filesystem tags
and mountable/writableflags...

Curt Sampson responded to der Mouse:

   Yes, it should. If the system can't support typed disklabels, it can't
   support them and that's that. That's no reason to take away the safety
   it provides on other machines that do support them.

So, what would be the ramification of those ports supporting a dual-disklabel
scheme?  Surely there's some blank space at the head of the disk where there
is room to write a BSD-style disk label...?

If it's a Sun disklabel, perhaps require a disklabel command to write the
Sun label as a BSD label at the second location.  I realise this goes against
the KISS principle, the principle of least astonishment and the principle
of unified namespaces, but it would allow the functionality shared by the
other ports.

And, actually, when I do a disklabel and change the type/cpg/fsiz/bsiz
fields on the label, they do remain in core, though they disappear if
I reboot ((Sarcasm) gee, I wonder why... (/Sarcasm))...

				--*greywolf;
--
Timesharing is dead.
Microsoft Windows-NT is going to take over the desktop completely.
UNIX will no longer exist.
And monkeys might fly out of my butt.