> Actually, I learned on a BSD system; however, I liked being able to
> tell what physical partition I was referring to absolutely; with BSD,
> I can't (so far as I can tell) refer to "the third partition, be it
> root, swap, ADOS, or regular".

Define "third".  If you mean "third in the usual list of eight
partitions each disk has", then you can indeed; that's the one with a c
at the end of the name.

If you mean "third in order from the beginning of the disk", that
depends on the disk label.  But...

> I can refer to the first root partition, the first swap partition,
> the whole disk, or the <n>th not root or swap partition.

...so does this; it's merely convention that the a partition begin at
the beginning of the disk, or that the b partition follow on its heels.
It's also purely convention that a is root and b swap; it's entirely
possible to configure a kernel "root on sd3f swap on sd2h".  Just
nobody would do such a thing under normal circumstances; conventions
are there because they tend to make things easier.

And the a and b partitions on non-boot disks get used for non-root and
non-swap filesystems (relatively) often.

(The whole-pack c partition is a special case.  I'm not sure how I feel
about having one; I definitely don't care for having it as the c
partition.  But that tradition is strong enough, and known by enough
code, that there's no way it'll change now.)

> I can't figure how you'd name the second swap partition.

sd1b, usually.  But it depends on what piece of disk you want to swap
on.  Set up the partition table so that some partition refers to that
piece of disk, then go ahead and swap on it...regardless of whether
it's a b partition.  At worst you'll be unorthodox and perhaps
confusing to someone who doesn't notice that you're swapping on an
unusual place.

> cNdNsN was at least an obvious identifier; it tells you which
> controller, which drive, and which partition, unambiguously.  I
> consider this a win.

I suppose it's a matter of taste.  I rather like having disk numbers
(the number after "sd") assigned from zero upwards as the kernel scans
the available controllers and target IDs.  This is occasionally useful;
for example, I know a machine whose boot disk - sd0 - is ID 1 and whose
second disk - sd1 - is ID 2.  Occasionally, I want to boot from the
second disk; I just change its ID to 0 and then it becomes sd0, with
the old sd0 becoming sd1.

Of course, it can also be a minor pain.  In another case, I would have
liked to have been able to nail down SCSI target 3 as sd0.  (I could
have done this by building a kernel with sd0 specifically set to target
3, but it was quicker and easier to re-jumper the other disk to an ID
higher than 3.)

> p.s.: I also liked that SCSI device 5 was device 5.  Always.  I don't
> like having sd0, cd0, and st0 - in fact, this confused me and made it
> difficult for me to install, because unit 5 was sd1.  ???

Again, it's a matter of what you're used to.  If you're used to the
disk at target 5 being sd5, anything else will confuse you and make it
harder for you to work; conversely, if you're used to the lowest target
disk being sd0, the next being sd1, etc, then again anything else will
confuse you and make it harder for you to work.

					der Mouse

			    mouse@collatz.mcrcim.mcgill.edu