On Friday 10 October 2003 08:52 pm, Christopher Browne wrote:
> Oops! nandrews@investsystems.co.uk ("Nigel J. Andrews") was seen
spray-painting on a wall:
> > I've not looked at many RPMs but I must say that the few I have have
> > never been relocatable. Can the postgresql RPMs not be made
> > relocatable?
> Unfortunately, relocation would have to include the init scripts, and
> that would be pretty hairy. The notion of "relocatable RPMs" came up
> early in its design, but the only case where that will be particularly
> usable is if the components are mostly binaries that only make
> relative path references. That situation is unusual, to say the
> least.
I've been watching this discussion with interest (well, I _am_ the RPM
maintainer, after all) and have to say that it has been thought of before.
It wasn't at that time implemented due to political factors (read: the then
Red Hat maintainer (@redhat.com) refused to include such support even if I
had built it). But I did go through the design phase. If everyone can be
patient, I'll try to go back into my archives and dig out the design doc I
put together way back then. In the meantime, I'd like to hear people's
ideas. As alternatives (debian-style) are fully supported in later Red Hat
(and the new Fedora Core) releases, a scheme that uses alternatives would be
ok.
Be sure to post to the pgsql-ports list instead of pgsql-general, though. If
the list server will accept it, reply-to has been set to pgsql-ports.
--
Lamar Owen
Director of Information Technology
Pisgah Astronomical Research Institute
1 PARI Drive
Rosman, NC 28772
(828)862-5554
www.pari.edu