Package EVR problems in Fedora 2008-06-10
Dominik 'Rathann' Mierzejewski
dominik at greysector.net
Fri Jun 13 08:23:43 UTC 2008
On Friday, 13 June 2008 at 10:15, Harald Hoyer wrote:
> Adam Jackson wrote:
> >On Wed, 2008-06-11 at 16:04 +0100, Richard W.M. Jones wrote:
> >>On Tue, Jun 10, 2008 at 12:24:36PM -0400, buildsys at fedoraproject.org
> >>wrote:
> >>>ocaml-deriving:
> >>> F8-updates > F9-updates (0:0.1.1a-4.fc8 > 0:0.1.1a-3.fc9)
> >>>
> >>>ocaml-gsl:
> >>> F8-updates > F9-updates (0:0.6.0-4.fc8 > 0:0.6.0-3.fc9)
> >>>
> >>>ocaml-json-static:
> >>> F8-updates > F9-updates (0:0.9.6-4.fc8 > 0:0.9.6-3.fc9)
> >>[etc etc]
> >>
> >>Is this wrong?
> >>
> >>I'm afraid to say that a lot of packages I have do this. The reason
> >>is that I develop and build packages on Rawhide, then backport them to
> >>F-8. However when backporting to F-8 I have to bump the release
> >>number up, typically because I have to add an ExcludeArch: ppc64[*]
> >>for F-8, but may be because of other packing twiddling too.
> >>
> >>I wasn't aware that there had to be a strict increase in package
> >>numbering between branches. (In fact, I wasn't aware that Fedora even
> >>allowed updating between Fedora releases).
> >
> >It's very strongly encouraged. We do provide upgrade paths between
> >releases (and are even working to make them more robust). So yes,
> >please do keep EVRs for older releases lower (in the rpmvercmp sense)
> >than those for newer releases.
> >
> >When in doubt:
> >
> >% sudo yum -y install rpmdevtools
> >% rpmdev-vercmp 0:0.9.6-4.fc8 0:0.9.6-3.fc9
> >0:0.9.6-4.fc8 is newer
> >
> >- ajax
> >
>
> this can be prevented automatically by a cvs-commit check script
The harm begins when you do `make tag', so I'd suggest having this
check there instead.
Regards,
R.
--
Fedora http://fedoraproject.org/wiki/User:Rathann
Livna http://rpm.livna.org | MPlayer http://mplayerhq.hu
"Faith manages."
-- Delenn to Lennier in Babylon 5:"Confessions and Lamentations"
More information about the fedora-devel-list
mailing list