Plan for tomorrows (20070418) FESCO meeting

Michael Schwendt bugs.michael at gmx.net
Thu Apr 19 08:47:50 UTC 2007


On Thu, 19 Apr 2007 07:10:26 +0200, Thorsten Leemhuis wrote:

> 
> 
> On 18.04.2007 23:05, Brian Pepple wrote:
>  >
> > /topic FESCO-Meeting -- MISC -- Vote on rebuilding packages with old
> > disttag - jwb 
> >
> 
> Something related to this (mentioned on FAB-list 24 hours ago, but 
> didn't get any replies) to fix the problem in F8 and later:
> 
> Why don't we just expand the disttag to something else without the
> "fc<I>n</I>" in it in the devel branch? Then we don't have run into the 
> "package disttag doesn't match the release it is shipped in" problem 
> when a package is not rebuild during a devel cycle. A simple ".1" maybe 
> -- that should make everybody happy afaics:
> 
> [thl at thl tmp]$ # this is how we do it right now:
> [thl at thl tmp]$ rpmdev-vercmp 0 1.0 5.fc6 0 1.0 5.fc7
> 0:1.0-5.fc7 is newer
> [thl at thl tmp]$ # why not do it like this:
> [thl at thl tmp]$ rpmdev-vercmp 0 1.0 5.fc6 0 1.0 5.1
> 0:1.0-5.1 is newer
> 
> Or am I or rpmdev-vercmp missing something here?

You are missing the rebuild.

Assume we create the FC-7 branch from current devel and target Fedora 8 in
devel. Then devel starts with the packages that have a dist tag .fc7 (or
older) or no dist tag. All the packages need at least one rebuild to
change the package release field. Only packages, which are not rebuilt,
keep their old Release. But builds in devel would get a .fc8 dist tag, so
when rebuilding a package why jump to .1 instead of .fc8?

If you want to rebuild all packages with a changed %dist and without
modifying the spec files to bump Release, the plan is flawed. "1" > "f",
hence "1" is newer than all fcN dist tags, okay, but you only get an
upgrade path from Fedora <= 7 to "1", since for Fedora 9 you would need to
jump from "1" to "2" for any rebuilds to be newer than Fedora 8 again.

So, conclusively, this proposal only shoves the problem under the carpet. ;)




More information about the Fedora-maintainers mailing list