Le mardi 27 juin 2006 à 12:22 -0400, Jesse Keating a écrit : > On Tue, 2006-06-27 at 17:42 +0200, Nicolas Mailhot wrote: > > True. However my personal take on this is updates from the previous > > release should never have a NEVR higher than that of rawhide, because that > > will screw moving from previous to rawhide which is a valid use-case. > > > > So even if updates frome previous do not work with rawhide, they should > > never be applied by the package manager even if the old repo is still > > configured. If that is not the case this is a repo, not user problem > > Doesn't always work that way though. For example, a new upstream > version of Foo is released. The maintainer is able to get it built and > released for FC5 Updates. However FC6 (rawhide/development) has gcc > issues or some other issues preventing packages from making it out, so > the rawhide one remains an older version. This can VERY easily happen. Sure, but that's all part of rawhide ususual turnover. It may happen but it should not happen for long and certainly not wait till the next test release to be fixed. Generally speaking few people are mad enough to run rawhide just after a release (I'm part of the madmen) so the usual upgrade is FCX + some updates -> Rawhide. Which implies the upgrade path must be yumable easily (and yes I know we won't manage it all the time but if we don't even try it will never work at all). Thanksfully with the janitor scripts the Fedora infrastructure has been sprouting ensuring this all work should become less of a hurdle than in the past. -- Nicolas Mailhot
Attachment:
signature.asc
Description: Ceci est une partie de message =?ISO-8859-1?Q?num=E9riquement?= =?ISO-8859-1?Q?_sign=E9e?=
-- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
