Re: Use with Ximian RPMs?

On Tue, 2002-07-16 at 12:02, John HEARNS wrote:

> > Rebuilding the database won't help.  You have hit a bug in RPM.  When

'bug' may not be the best description here. How to sort such things is
TOTALLY arbitrary. They just used a different arbitrary rule than some
people would have.

> > presented with two versions to compare, one starting with a letter, the
> > other with a number, rpm gets confused and always returns -1, which
> > up2date takes as 'upgrade this package'.  So you get into a situation
> > where up2date always goes for the _other_ package if two are available
> > to it, e.g. pilot-link-0.9.5-8 and pilot-link-0.9.5-ximian.2.
> > 
> 

> Incidentally, I have come across a similar problem recently.
> No names, no pack drill (this is a public list)
> A package was upgraded - RPM renamed  from package-1.1.rpm to
> package-latest.rpm
> Current barfs on this also when trying to create a channel
> (it tries to find package-1.2.hdr for instance)
> No fault at all in current.
> 
> I'm no RPM expert at all, but I idly wonder if there is a better
> way than getting the version number from the file name.
> Guess this would involve some fundamental changes, so I'll
> keep quiet.

We don't. We use the rpm python bindings to get that information from
the rpm itself. The only thing we use the filename for is actually
finding the package on disk. 

What does rpm -qpi package-latest.rpm tell you?


-- 
Hunter Matthews                          Unix / Network Administrator
Office: BioScience 145/244               Duke Univ. Biology Department
Key: F0F88438 / FFB5 34C0 B350 99A4 BB02  9779 A5DB 8B09 F0F8 8438
Never take candy from strangers. Especially on the internet.