* Martin Pitt <mpitt@debian.org> [2008-10-10 09:49:01 CEST]:
> Alexander Wirt [2008-10-10 7:02 +0200]:
> > mechanized? No.
>
> I meant it in the sense of "run a script to create a backport from a
> particular testing/unstable release, as opposed to changing any source
> package and upload it manually to backports.org". I would very much
> assume that this is what currently happens with backports.org. At
> least that's how we do backports in Ubuntu, with "backport-source.py
> package_name source_release".
Erm, the source package _has_ to be changed, the version has to get
adapted and the likes, for a start, propably even build-dependencies.
And it's expected that people uploading their packages to backports
apply similar testing to their uploads than they do with uploads to
unstable.
> > Only if they are tested carefully.
>
> Goes without saying.
mechanized didn't sound like that, to be honest.
> > And I still don't like this.
>
> --verbose ?
,----------------------> quote yourself <----------------------
| So a compromise I can live with is to put it back into unstable (or
| even just experimental), but never let it propagate to testing. Then
| backports.org can do mechanized backports of updates without being
| tied to the long lifecycle of Lenny. Would that be an acceptable
| compromise for all involved parties?
`----------------------> quote yourself <----------------------
Backports are meant to sit between stable and testing so that people
can upgrade to the next stable release without any major headaches. If
you backport from "unstable (or even just experimental)" you lose this
approach totally and fail with what backports.org is trying to achieve.
This is what formorer doesn't like, and honestly, as much as I would
like to help getting things working again and support postgres users
here, I have to agree with him.
So long,
Rhonda