Re: createrepo -c problems
John Newbigin <[email protected]>
| Newsgroups | gmane.linux.rpm.metadata |
|---|---|
| Message-ID | <[email protected]> |
Jay Soffian wrote: >> I thought I had merged those patches. > > > I think I figured out why. > > Hans-Peter's offers two patches to fix the missing hdrid here: > > https://lists.dulug.duke.edu/pipermail/rpm-metadata/2005-July/ 000538.html > > I don't see a reply to that message. > > A virtually identical change to one of those patches was offered by > Christoph Thiel here: > > https://lists.dulug.duke.edu/pipermail/rpm-metadata/2005-November/ > 000552.html > > You applied Christoph's patch which is equivalent to the "less- robust" > of Hans-Peter's offerings. > > In any case, I want to revisit the original question: > >> Perhaps if there is no useful id/sum then use the version-release >> strings? I have modified my createrepo to use csumtag = '%s-%s-%s-%s' % (self.hdr['name'], self.hdr['version'], self.hdr['release'] , self.hdr['hdrid']) If there is an ID (you could use SHA1) then good, if not it still does not break because of the NVR. NVR is not always unique because you could have multi ARCH or EPOC, and perhaps others. Hans-Peter's createrepo-nohdrid-take2 looks OK but I think for simplicity, the same hash should be used for all packages. I did a quick check and all the packages I have, have an MD5. John. > > > It's not clear to me what's wrong with basing the cache filename on > NVR. These are supposed to be unique in the RPM universe (at least > within a distribution) since if they aren't worlds collide. > > j. > _______________________________________________ > Rpm-metadata mailing list > [email protected] > https://lists.dulug.duke.edu/mailman/listinfo/rpm-metadata > > > -- John Newbigin Computer Systems Officer Faculty of Information and Communication Technologies Swinburne University of Technology Melbourne, Australia http://www.ict.swin.edu.au/staff/jnewbigin