[vanco@sonic.net: Re: dpkg modification: non-interactivity]
Wichert, you're right, I didn't mean to send this to only you ;)
Forwarding...
----- Forwarded message from Aaron Van Couwenberghe <vanco@sonic.net> -----
Date: Wed, 6 Jan 1999 03:47:06 -0800
From: Aaron Van Couwenberghe <vanco@sonic.net>
To: Wichert Akkerman <wakkerma@cs.leidenuniv.nl>
Subject: Re: dpkg modification: non-interactivity
X-Mailer: Mutt 0.93i
On Wed, Jan 06, 1999 at 06:01:50PM +0100, Wichert Akkerman wrote:
> Previously Mitch Blevins wrote:
> > I just think we can accomodate the needs of the multi-machine clusterer's
> > without adding alot of complexity to the tools used by everyone else.
>
> But making a filesystem appear to change while in reality it doesn't
> is not the solutition here. I still think the only way to do this
> properly is make the maintainer scripts check this.
>
> Wichert.
>
here you will be giving an automatic bug to everyone who doesn't fix his
pre/post inst/rm. Assuming standard config directories exist ( /etc )
locally and that the packages conform to policy, having a fake filesystem in
place is the least intrusive way.
Except, when a pre/post inst/rm depends on a file it generated from
something (ie an idl file, or some intermediate form like yacc or m4....)
being there, then you break it. So you're forced to make some kind of
overlay filesystem that keeps changes in a separate cluster in memory, and
overrides the true FS with the dummy one...
And then you're stuck in a tar baby. Wichert, I agree with you. I just don't
think it should be done _either_ way. What I'm saying is I don't think
admintool should have to deal with this very uncommon case; it will burden
debian as a whole too much.
--
..Aaron Van Couwenberghe... ..vanco@sonic.net.. ..aaronv@debian.org....
Berlin: http://www.berlin-consortium.org
Debian GNU/Linux: http://www.debian.org
----
Nullum magnum ingenium sine mixtura dementiae fuit.
[There is no great genius without some touch of madness.]
-- Seneca
----- End forwarded message -----
Reply to: