Re: db1/glibc debacle
On Thu, Aug 08, 2002 at 10:14:43AM -0400, Ben Collins wrote:
> For those of you wondering if this was intentional or not, consider it
Breaking stuff intentionally is bad, mmmkay?
> So here's a quick FAQ:
Q. Which packages get broken when you update libc?
Q. Is this enforced using the dependency information?
> Q. You're a bastard and you should have given this more thought by
> checking for which packages used this.
> A. Let's stop with the name calling. First off, it's near impossible
> (without unpacking everything and running objdump on it) to tell what
> uses libdb.so.2, since it is included with libc6.
If unpacking everything and running objdump is the way to go, that's
the way to go. At the very least libc6 should Conflict: with all the
packages it breaks outright to help make sure people's systems don't
unpredictably break when people do partial upgrades.
> Q. You dip shit, I have to have this working on my mission critical
> applications and systems. You broke all my servers and my boss is
> threatening to fire me! Are you going to make up the difference in my
> salary now?!
> A. Silly rabbit, unstable's for developers.
First, that's not particularly correct: unstable's for whoever is happy to
deal with a fair degree of breakage, not just developers.
Second, *testing* doesn't have this limitation, and is expected to not
have major breakages and is expected to be used by users.
We need to fix this problem (probably by running objdump and making
the new libc6 have a versioned Conflict: with any packages from
woody/sarge/sid that need libdb1), not tell everyone who's system breaks
to FAQ off...
Anthony Towns <firstname.lastname@example.org> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. GPG signed mail preferred.
``If you don't do it now, you'll be one year older when you do.''