Broken upgrade paths in F7
Patrice Dumas
pertusus at free.fr
Tue May 15 17:11:34 UTC 2007
On Tue, May 15, 2007 at 11:13:07AM -0400, Jesse Keating wrote:
>
> To be perfectly honest, I'm not completely happy with this process for the
> final freeze either. This was an experiment for the Fedora 7 release, and it
> went really well for the test release freezes. I agree that the final freeze
> should be done differently, however I didn't want to change it at this point.
Agreed. I think it should be changed for the next release, it is to late
to change it now. But I wanted to voice out my concern so that things
are changed next time. I already complained at the first time the release
process was outlined but nobody cared. (As a side note, for the freeze
there was still plague used and we hadn't to ask to push so of course it
was better than for the release.)
> Asking how many were rejected isn't quite fair. It's not the people who are
> asking for tagging that I'm concerned with, it's the people who just don't
> realize we're in a freeze and happily drop in a soname bump that breaks a ton
That shouldn't happen if instead of make build, one have to do
make release-build
when we are in freeze.
> of packages or other such acts without noticing that we're in a freeze and
> doing such can cause problems with the release. This has happened in the
> past, I'm not just being overly paranoid. I very much want there to be less
> overhead, and I welcome any effort into the workflow that gets us there, but
> at the same time we have to be able to manage a freeze and have control over
> what gets in and what doesn't.
That's why I think that rel-eng people could disallow a push and should
have the last word, at least until the packager had time to say why he
wants it. That's what was done for extras and it worked well -- although
I agree that how it was done in extras is not necessarily suitable for
merged fedora, especially because of the constraints linked with
composing the release.
--
Pat
More information about the Fedora-maintainers
mailing list