Subject: Re: Proposed rc.d changes....
To: None <tech-userlevel@netbsd.org>
From: Greg A. Woods <woods@weird.com>
List: tech-userlevel
Date: 05/01/2000 12:43:41
[ On , May 1, 2000 at 10:21:26 (-0400), Nathan J. Williams wrote: ]
> Subject: Re: Proposed rc.d changes....
>
> <lukem@cs.rmit.edu.au> (Luke Mewburn) writes:
>
> > 3. Reworking the configuration mechanism (replacing rc.conf)
> ...
> > My proposal is to replace rc.conf with separate
> > /etc/rc.conf.d/foo config files, one for each service, with
> > contents similar to:
> > foo=NO
> > foo_flags="-q ex0 -p 667"
>
> It should be specified whether these are really going to be evaluated
> by /bin/sh, or just interpreted as a set of name=value pairs, one per
> line, with sh-like quoting (the latter does not preclude the
> implentation from using /bin/sh to interpret it, of course).
I would hope that the definition is that these config variables are
simply "name=value" pairs with simple quoting. I wouldn't even go so
far as to allow the the quoting to be "sh-like" (I can't imagine any
reason to use some of the more complex forms of shell quoting!).
Unfortunately this does mean that a real-world implementation MUST avoid
using /bin/sh to parse these variable settings. Assuming that the
syntax definition is restricted for the express purpose of making it
possible to parse these variable settings in other ways that totally
avoid using /bin/sh, any use of /bin/sh directly would present just too
many opportunities for users to do undefined things.
--
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>