[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: slink -> potato

Anthony Towns <aj@azure.humbug.org.au> wrote:
> Hmmm. I can't actually find any mention of this in policy. In fact,
> discussion of what should be done when in prerm and postrm seems pretty
> bare, period.

OK, so it's not actually in the policy.

> What sequence of events is actually going to cause problems? I'd have
> thought dpkg would generally manage to keep dependencies pretty reliable
> while upgrading.

The idea is that when you upgrade the package like telnetd, there may be
new shlib dependencies, etc. which means that you should stop spawning
new daemons until it is configured.  Of course, this may not happen for
every release, but the prerm file comes from the old version, so it can't
tell whether this is necessary, but it is the only one that knows exactly
how to stop the daemon from spawning, so it just has to stop it every time.

This applies to standalone daemons as well in the case when the have
configuration files which may change in an incompatible way.

One thing that dpkg doesn't do (AFAIK) is to deconfigure a package when
something it depends on is deconfigured, upgraded or removed.  If this were
the case, then the above would work even better.

> Did anything come of the `locally-essential' thing that apt was going
> to support at one point? So that you could say "telnetd is Essential
> to this machine --- I do remote maintenance", and have Apt configure
> telnetd (and any dependents) ASAP. That would solve the original problem
> pretty well, it seems to me.

Anyway, back to the original problem, the best solution IMHO is just to
run telnet/ssh and screen so that none of this really affects you.
Debian GNU/Linux 2.1 is out! ( http://www.debian.org/ )
Email:  Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt

Reply to: