On Tue, Jun 20, 2000 at 11:00:59PM +0100, Julian Gilbey wrote:
> (2) The sysadmin upgrades the foo package. Dpkg notices that
> /etc/foo.conf has changed and offers to upgrade it. The sysadmin
> agrees. Now what happened seems to be that after this (and I
> don't quite know how this works exactly) /etc/foo.conf is mv'ed to
> /etc/foo.conf.dpkg-old, /var/pkg/conf remains a hard link to it,
> that is to /etc/foo.conf.dpkg-old, and /etc/foo.conf is the new
> conffile. But now the /var/pkg/conf file is the old conffile and
> /etc/foo.conf is the new one. Oops.
Ah, I see what you're saying. You're talking about the case where the
package actually *contains* the hard link, rather than, say, making it
in the post-inst. Yes, that will definitely break badly.
In that case, I suggest changing the first line of the proposal from:
A package may not make hard links to conffiles.
to:
A package may not contain hard links to conffiles.
I still support the proposal, with or without my proposed change, but
I do think it's a little more clear as I phrased it.
cheers
--
Chris Waters [EMAIL PROTECTED] | I have a truly elegant proof of the
or [EMAIL PROTECTED] | above, but it is too long to fit into
| this .signature file.
