Subject: Re: `rc.local.conf': bad name
To: NetBSD Userlevel Technical Discussion List <tech-userlevel@netbsd.org>
From: Greg A. Woods <woods@weird.com>
List: tech-userlevel
Date: 07/31/2000 19:14:31
[ On Monday, July 31, 2000 at 14:03:47 (-0700), Greywolf wrote: ]
> Subject: Re: `rc.local.conf': bad name
>
> On Mon, 31 Jul 2000, Greg A. Woods wrote:
> 
> # > # and the idea of gettting rid of (/etc/rc.conf) completely,
> # > # incorporating the defaults directly into the various scripts.
> # > 
> # > Kill that before it multiplies.
> # 
> # If you had an original copy with which you can do a three-way diff,
> # either in RCS or /var/backups, why would you care?
> 
> Because I think the information should be centralized.  My objection is
> not over being able to do diffs, or the virtues of source control;
> my objection is to splintering all the params out into their rc scripts
> (opinions previously stated on (can_of_worms (*(**)())[])_that_).

Ah, Ummm..., either you didn't read what I wrote; or I didn't write what
I meant to write; or you really do need to rethink your position from
the ground up.

I'm *NOT* talking about creating a whole bunch of new little config
files, but rather quite the opposite.  I'm talking about moving the
configuration *defaults* into the same places where they are used and
maintained, thus making the developer's jobs *MUCH* easier, while at the
same time keeping local changes to those defaults in one central master
file -- a central file that should contain only localisations.

> # People really should learn to use the likes of RCS (or SCCS! :-)
> 
> RCS was invented because AT&T was tight-fisted about SCCS.

Ah, no, not really.  Maybe it was because of drawbacks perceived in
SCCS, though as the "Applying RCS and SCCS" authors say (paraphrased),
since RCS was created a full decade after SCCS and in full view of SCCS,
why isn't it a whole lot better than it is?

Given that the BitKeeper guys chose to stick with SCCS even after having
lots of experience with RCS, and after doing at least some technical
analysis of RCS, one has to even further wonder at the merits of RCS.

(BTW, tight-fisted or not, there are now at least two free
implementations of SCCS! ;-)

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>