Subject: Re: /etc/inittab (was: /etc/default)
To: Captech) < (James Graham>
From: Greg A. Woods <>
List: current-users
Date: 07/27/1995 16:46:12
[ On Wed, July 26, 1995 at 11:35:53 (PDT), James Graham wrote: ]
> Subject: Re: /etc/default
> 	Secondly, in case nobody has noticed, "init s" doesn't always do
> 	the right thing -- at least it hasn't for me.  Sometimes it will
> 	go into single-user mode (or so it says) and not kill quite a few
> 	of the networking processes.

Although it's not well documented (that's what happens when you try to
read between the lines in a UNIX manual page), 'init s' is intented only
for use from boot up, or internally from /etc/rc1.  You should never run
it from the command-line, as it usually does nothing more than change
the internal state of init.
> * Init run states
> 	The idea of 8 run levels (S0123456) is excessive, especially con-
> 	sidering that 4 of them are used for shutting the system down.

Personally I'd extend it further to perhaps as many as 256 states.
There's no telling what some bright young person who simply finds a
feature like this (but knows none of the political crap surrounding it)
might implement....

> 	Berkeley had the right idea (at first, anyway):  You can really
> 	only be in one of the following states:

Well, on a simple general-purpose UNIX server, yess, that's about it.

As I say, there's not telling how far you could take this in specialized
applications domains.

> 	When I hit ^D, I want to *walk away* and not have to look back.

Well, now you know you've got to wait a moment and be a bit more patient.

> And, of course, being a BSD die-hard from 4.2 (some of you come from an
> even earlier age, I'm sure), I don't see SVR4isms as being a win in this
> arena.

I personally come from the V7 era, but have used the SVRx'isms to great
advantage in many different situations....

							Greg A. Woods

+1 416 443-1734			VE3TCP			robohack!woods
Planix, Inc. <>; Secrets Of The Weird <>