On Sun, 06 Dec 2009 at 05:04:49 +0100, Cyril Brulebois wrote:
> Russ Allbery <rra@debian.org> (20/08/2009):
> > I think we only want to do this check if the first line of the
> > Changes file says UNRELEASED, since there are valid use cases for a
> > mismatch otherwise.
>
> Personally, I would have *really* ***loved*** to have such a check in
> place. #559659 led me to upload a few dozens of packages to the wrong
> distribution, without having a fighting chance since lintian said
> nothing.
>
> As a “compromise” with what you were proposing, I guess it could be an
> E: in the UNRELEASED case, and a W: in other cases?
Having managed to upload an experimental package to unstable *again*, I've
written a Lintian check for this. It specifically looks for an
unstable/experimental mismatch and reports it as important/certain;
other apparent mismatches are normal/possible, because Russ said
"there are valid use cases" (I'm not sure what they are, perhaps you could
add them to the .desc?). Adding a specific check for UNRELEASED would
probably be good too.
The parsing is a bit suspicious because I couldn't work out how to make
Lintian cross-reference debian/changelog with the changes file, so I'm
just using the Changes field and a regex. It's better than nothing, though...
Simon
Attachment:
signature.asc
Description: Digital signature