Beyond that,
figuring out dependencies is a mess in Redhat. Probably is for
all distributions. But it's the main reason I use PlanetCCRMA instead of
doing it all by hand. Any solution to that problem?
PlanetCCRMA is a great idea and a great tool, but its rpms are quite
generic, so you may have problems if they are dynamically linked to old
libraries (glibc most notably). That's why it's best to use rpms
designed specifically for your distro AND your version of that distro.
Could you explain in more detail what would be the difference between a
"generic" and a more distribution specialized rpm? For rpms I package
myself for Planet CCRMA I create spec files that follows (as much as I
can) RH conventions. Generic spec files, usually included in the
packages themselves, are most of the time heavily revised and edited to
match RH as well, and the same thing happens to spec files coming
originally from other distributions.
Regarding dynamic linking, I don't quite see what the problem might be.
Maybe you are not aware that Planet CCRMA rpms currently come in three
"flavors", 7.2, 7,3 and 8.0, each one being a separate repository. Each
program has three versions of the binary rpm(s), one for each distro
version, each one compiled under the proper version of the distribution,
with all current updates. As long as RH itself does not break backwards
compatibility within a given revision with its updates there should not
be any problems whatsoever with dynamic linking.
-- Fernando