Subject: Re: post-installation and rc.d enhancements
To: Frederick Bruckman <email@example.com>
From: Bill Studenmund <firstname.lastname@example.org>
Date: 04/18/2002 10:40:53
On Thu, 18 Apr 2002, Frederick Bruckman wrote:
> On Thu, 18 Apr 2002, Alan Barrett wrote:
> > On Wed, 17 Apr 2002, Frederick Bruckman wrote:
> > > we don't have to edit the file -- we can just re-install the whole file
> > > with each package that has an rc.d script.
> > No! Please don't do that!
> > People keep old binary packages lying around, and expect to be able to
> > reinstall them later (perhaps to downgrade after a bad experience with
> > a newer version of the package). The old binary package would not know
> > about the new variables in the monolithic config file.
> Huh? Why would that matter. If an old package rc.d script has no
> knowledge of a particular variable, it'll just ignore it. Except for a
> few odd corners, you're mostly only going to have a single variable
> that matches the name of the script. I'm only talking about a
> monolithic "defaults" file -- the user is expected to copy & paste
> from there to "rc.conf".
More to the point if we have a monolithic config file for packages, adding
an "old" package would install the older monolithic config file, no? Thus
_reverting_ whatever advancements other packages may need. ?? Or am I not
understanding what you propose?
The only way I can see to make package rc.d scripts work is if all of the
defaults script foo needs are in files that come only with package foo
(other than the global "YES" that is needed to turn script foo on).
Anything else and we open ourselves up to a migration nightmare.
Or is that not what you're proposing with a monolithic config file?