Re: Recovering /var (package status only)
On Thu, Jan 23, 2003 at 03:03:07AM +0000, Karsten M. Self wrote:
> on Wed, Jan 22, 2003 at 12:22:47PM -0800, Osamu Aoki (email@example.com) wrote:
> > On Wed, Jan 22, 2003 at 10:00:28AM +0000, Colin Watson wrote:
> > > On Wed, Jan 22, 2003 at 01:03:00AM -0500, Travis Crump wrote:
> > > > Question about false positives(/usr/share/doc/ directories that don't
> > > > correspond to a package): Are they a bug or is there nothing
> > > > wrong with them? On my system I have the following false
> > > > positives:
> > > >
> > > > debian-reference-en, debian-reference-common: /usr/share/doc/Debian
> > > > doc-linux-text: /usr/share/doc/FAQ
> > > > doc-linux-text: /usr/share/doc/HOWTO
> > > > doc-debian: /usr/share/doc/debian
> > >
> > > doc-linux-text and doc-debian are special cases, probably
> > > debian-reference-* too. I tried to formalize the upper-case rule for
> > > doc-linux-* on debian-policy in August 2000, but never got round to
> > > following through on the approving noises I got in response.
> > In DDP, we discussed to move all DDP document into /usr/share/doc/Debian
> > just like HOWTO and others. It stand out when browsed by "mc" :)
> I'd support this on an organizational basis. These aren't packages,
> they're additional documentation concerning a specific topic. This
> should also harmonize (evil in copyright, good in directory names) the
> "lowercase is packages" association.
> > organization but now we have very compelling argument for disaster
> > recovery :)
> I prefer to think of this as a validation of consistency, rather than a
> feature which should be explicitly aimed for. There's a distinction.
> Admins *shouldn't* be blowing away their /var partitions, and they
> *should* be backing up critical system data, preferably with multiple
> offsite backups. Yes, we know this isn't the case, but does Debian
> policy need to accommodate poor practices?
BTW, admins should keep /var/backup/* as separate partition if they care.
This is THE BACKUP of critical system package state. The
/usr/share/doc/[a-z][\+\-_a-z0-9:]* are only the 2nd stage backup.
> From a consistency viewpoint, however, it *does* become a useful audit
> check to see if there are any non-package directories consisting of all
> lowercase alpha characters in /usr/share/doc. The fortuitous
> consequence that you now have a backup representation of package state
> is useful. It's not an intentional result. Subtle but important point,
> namely: if policy at some future point dictates that this arrangement
> should change, then disrupting the "but you're destroying our backup
> dpkg status state repository" argument holds no water.
You may be right. But if one proposes to disrupt this very reasonable
policy, one needs to have very compelling reason. (Especially when
people can create arbitrary directory entries without violating policy as
long as they put one upper case.)
I think this is a policy item with "should" phrases.
~\^o^/~~~ ~\^.^/~~~ ~\^*^/~~~ ~\^_^/~~~ ~\^+^/~~~ ~\^:^/~~~ ~\^v^/~~~ +++++
Osamu Aoki <firstname.lastname@example.org> Cupertino CA USA, GPG-key: A8061F32
.''`. Debian Reference: post-installation user's guide for non-developers
: :' : http://qref.sf.net and http://people.debian.org/~osamu
`. `' "Our Priorities are Our Users and Free Software" --- Social Contract