Re: RPM should be renamed to WTF

On Tue, 14 Oct 2003, Eric G Ortego wrote:

> Lars Damerow wrote:
> 
> >>From Eric G Ortego <[email protected]>, Tue, Oct 14, 2003 at 08:56:05AM -0500:
> >  
> >
> >>dependancys. Ironicly this is exactly what I hate about rpm's. Why does 
> >>a source package have to depend on other packages?
> >>    
> >>
> >
> >Simple--you often need other libraries to be present in order for a compile to
> >work.
> >
> I see.  But that shouldnt differ from tgz's should it?  Or are most 
> rpm's binary packages?

RPMs named .i386/.i486/.586/.i686/.athlon/.whatever/.rpm are generally
binary RPMs. It's quite common to see these requiring libfoo-1.2.3 as a
dependency, because that's the libraries that the binaries they contain were
compiled and linked against (and thus, the only version *guaranteed* by the
packager to work with the binary package they've assembled). RPM does have
the facility to specify libfoo-1.2.3-or-greater and libfoo-1.9.9-or-older,
but it's upto the packager to use it, and use it correctly.

RPMs named .noarch.rpm are generally perl/python/whatever scripts, and thus
portable across architectures. It's quite common for such packages to
require the perl/python interpreters, since they're useless without the
appropriate interpreter.

RPMs named .src.rpm are what the previous two categories of packages are
built from. It's quite common to see BuildRequires dependencies, as the
packager may have found that whilst gcc-2.7.2 produces buggy object code,
gcc-3.2+ generates as-expected object code.

> And if an rpm is a source package I don't see how rpm improves anything.

RPM's dependency features warn you that if you install/rebuild a package,
things may not work as expected afterwards. It's always possible to override
the warnings if you (think you) know better, but the dependencies are
usually there for a good reason. If you're lucky, you may not find that
reason.

> >cheers,
> >lars
> >pixar animation studios

Heh - just caught Finding Nemo this afternoon. Nice work!

Alex.
-- 
Alex Butcher      Brainbench MVP for Internet Security: www.brainbench.com
Bristol, UK                      Need reliable and secure network systems?
PGP/GnuPG ID:0x271fd950                         <http://www.assursys.com/>