Subject: Re: Another changer, another changer problem
To: Curt Sampson <email@example.com>
From: None <firstname.lastname@example.org>
Date: 10/11/1998 11:07:16
In message <Pine.NEB.email@example.com>, Curt S
>I coded up a bit of proof-of-concept stuff for the new NetBSD
>startup dependency system that Perry and Matt were doing some work
>on. This made the whole concept of `runlevels' obselete by instead
>allowing you to turn on and off named daemons or named groups of
>daemons. So you wouldn't for example, have to have your radius
>server running just to run xdm on that machine as well.
Oh, that's very nice. That does sound like a vast improvement on runlevels.
(And, if you want, you can always implement runlevels in terms of it.)
>The problem with this is it adds a lot of mess to the system for
>very little return in functionality on the vast majority of machines.
I think a proper sparse device filesystem wouldn't add much mess; it could
be another optional filesystem like /procfs and /kernfs, and if you don't
use it, you don't pay.