From: jwinstea@jarthur.claremont.edu (Jim Winstead Jr.) Subject: Re: Device names (was Re: ttys2 not responding) Date: 8 Jul 1992 01:08:16 GMT
In article <5296@mccuts.uts.mcc.ac.uk> LeBlanc@mcc.ac.uk writes:
>The ttys? discussion was a tug of war between the Unixers (who like
>devices with 0 origin) and the DOSsers (who like compatibility with
>MS-DOS). On this question it seemed to me that the Unixers had
>better arguments.
Er, yes, I think I was the one who argued for zero-based stuff,
actually. :) There are, of course, compelling arguments for
ttys[0-3] and ttys[1-4]. Personally, I use the ttys[1-4] (and that's
what's on the root disk), 'cause that's how I've done it for some
time. My main concern in the naming of the devices is that they be
consistent, which I can see ttys[0-3] being more so than ttys[1-4].
Watch out for change on the next root disk.
>The fd* discussion had much less unanimity. I was myself fed up
>with the Minix device names, which I could never remember, and
>decided to adopt the attractive proposal made by Linus himself at
>nearly the end of the discussion: /dev/fd0H1440, for example, means
>floppy disk drive 0 (opposed to 1), with high density (opposed to
>Q or D or S) on a 3 inch drive (H-Q-D-S for 3 inch drives, h-q-d-s
>for 5.25 inch drives) and 1440 kilobytes accessible (instead of
>1200, 720, 360, etc).
This does seem to be the most reasonable solution I've seen (certainly
much better than some I've seen suggested), and is definitely both
more extensible and consistent than the stupid at[01] ps[01] PS[01]
things.
>There was also a discussion of the names for SCSI devices. In the
>latest MCC version I have used sda, sda1..sda8, sdb, sdb1..sdb8,...,
>..sde8; these were one of the suggestions made at the time, and
>conform to the hda/hdb pattern. There were also suggestions for
>distinguishing between controller cards and other complications.
I agree with this completely. It makes sense, and it fits with the
/dev/hd[ab]* stuff. Drew has suggested using /dev/sd[01][a-z] before,
but I think that keeping it similar to the /dev/hd* stuff is better.
(And it makes the MAKEDEV script easier to write. :)
>As far as I know, the scheme I have adopted has only one disadvantage:
>many people don't like it. No definitive solution can be found in
>the documents in /pub/Linux/standards at banjo.concert.net, but the
>other discussions are in mail-archives in the same directory.
Having people like it would be nice, but you can't please everyone.
Cumbersome names are absolutely silly, and I think the choices you've
made are completely reasonable.
--
+ Jim Winstead Jr. (CSci '95)
| Harvey Mudd College
| jwinstea@jarthur.Claremont.EDU
+ This is all my words. Honest!