Re: problems in upgrade
On Wed, Jun 27, 2001 at 02:04:22AM +1000, Anthony Towns wrote:
> On Tue, Jun 26, 2001 at 11:20:57AM -0400, Ben Collins wrote:
> > On Tue, Jun 26, 2001 at 05:13:17PM +0200, Santiago Vila wrote:
> > > I wonder why woody's libc6 depends on libdb2 in the first place.
> > > What is the exact nature of this dependency?
> > I'm looking into a means for ensuring upgrades wiht this setup, but I'm
> > open to suggestions (and moving a hacked libdb.so.3 back to libc6 is not
> > an option I want to work with).
>
> You might have to...
>
> Hrm. libdb2 isn't essential: yes, so everything that uses it has to depend
> or pre-depend on it. But they might also depend or pre-depend on libc6,
> and assume that'll imply a functional libdb2.
The libdb.so.3 in libdb2, that is historic from libc6 2.1.x, is not
going to be used by newly compiled programs (those compiled against
libc6 2.2, as opposed to libc6 2.1). Libraries cannot be essential, so
the solution here is to somehow make sure that libdb2 gets upgraded
along with libc6. It has to be "unpack libc6, then unpack libdb2". The
two do not have to be configured (and the order can actually be
reversed, so long as nothing is attempted to be configured during the
two being unpacked).
> If you could manage to get a libdb2 that worked with the old libc6, and you
> could set up something like:
We could always do what redhat has done, and copy the libdb.so.3 from
the glibc 2.1.3 binaries, and use them in the libdb2 package. That means
it will work when 2.1.3 is still installed, and before libc6 2.2.x is
unpacked. The solution will work, since we don't need libdb.so.3
compatibility for new archs (since they have never compiled anything
against it).
I loathe this solution, but if it's prefered, I'll implment it.
--
-----------=======-=-======-=========-----------=====------------=-=------
/ Ben Collins -- ...on that fantastic voyage... -- Debian GNU/Linux \
` bcollins@debian.org -- bcollins@openldap.org -- bcollins@linux.com '
`---=========------=======-------------=-=-----=-===-======-------=--=---'
Reply to: