Re: devfs

Michael Neuffer (
Sun, 18 Jan 1998 10:59:31 +0100 (MET)

On Sun, 18 Jan 1998, Richard Gooch wrote:
> > While I perfer the more abstract device/subdevice concept to the
> > more specific device/partition/slice proposal, device/subdevice is
> > not really general enough to handle slices within partitions without
> > using a contrived subdevice encoding. I don't have a reasonable
> > suggestion to offer for extending it ("k" for contrived, anyone?
> > :-)), so it looks like using device/partition/slice is the best
> > we're going to do.
> OK. Hopefully this covers everything. So, we have:
> /dev/sd/c0b1t2d3p4 for a whole partition, be it primary or logical
> /dev/sd/c0b1t2d3p4s5 for a single slice in a partition
> Leonard: one last choice before I set this in stone: do we use 'd' for
> LUN or 'u'? Since we have to drop the device/subdevice concept, I
> suppose we could reconsider the d/u choice too. IMHO 'u' is a little
> more obvious, although I'll leave it at 'd' if you really think that's
> better.

I think 'u' would be the better choice since it is more abstract.

'd' leads one to think of disks and that might create confsion more
easily for new users ("This is my second disk in the scsi chain so maybe I
need to write /dev/sd/c0b1t0d1 ?")