Subject: Re: tty device names for Cyclades Z
To: Chris G. Demetriou <firstname.lastname@example.org>
From: Nathan J. Williams <nathanw@MIT.EDU>
Date: 06/02/2000 21:57:44
<email@example.com> (Chris G. Demetriou) writes:
> > I don't care that my SCSI disks are attached to adv*, adw*, ahc*, ncr*,
> > siop*, or whatever. They're sd*. Why should ttys be different?
> And, if you're gonna go there (and i'm not saying we should 8-)...
> why should ethernet interfaces be different?
I buy this. It's good for support purposes, in that it makes the
configuration of simple systems much easier to guess. "you have one
ethernet card? ifconfig eth0."
It also simplifies install goop somewhat, again for the common, simple
systems. Being able to usefully guess eth0 is much nicer than:
ifconfig -l | sed -e 's/lo.//g;s/sl.//g;s/ppp.//g;s/strip.//g;s/tun.//g'`
which worked in 1.3.2, but was ugly and fragile (I realize that we now
have ifconfig -lb to accomplish this goal, but I think the point still
> actually, there's probably an argument that read/write disks,
> i.e. 'sd' and friends, should be indistinguishable from IDE (and
> perhaps other) disks, too...
Though you then need to integreate the various frobbing tools for
different back-ends (scsictl, atactl), along with some way to figure
out what the appropriate set of frobs is. We don't have good prior art
here (see the disaster of tools that is *ctl and *config).