On Fri, Apr 04, 2025 at 07:02:06AM +0000, Simon Quigley wrote:
> I can think of one specific case where an exception to the rule would be warranted, and it is somewhat rare.
>
> Sometimes, we need to introduce a brand-new source package late in a cycle, for a good reason. Take the extreme case where the Debian NEW queue is like, 1000 packages. Rust, or something.
>
> That package is uploaded to both Ubuntu and Debian at the same time. The uploader states their good reason, an AA does a full review, it gets accepted with a lower version number (0ubuntu1).
>
> The package is then accepted into Debian at some later point, with the exact same (de-facto) maintainer. For the sake of argument, the uploader is both a Debian Developer and an Ubuntu Core Developer. It would make sense to drive that through in Debian most of the time.
This is a good example of an edge case. Thanks!
If I may draw a clear line around it, I suggest that this is the case
where a Debian upload with the same source package name is "locked" to a
person or team through an an existing ITP bug in Debian, and the Ubuntu
uploader is the same person or team.
If this is a short term thing, then I don't think that adding something
to debian/README.source is even worth it. It might be a good idea to
define some recommendations around package version strings to use in
this case, but I think that's best deferred as it seems more scope creep
than needed for this particular thread. I'd be happy just to treat this
case (as I've tried to define in the previous paragraph) as out of the
scope of the general recommendation to always use `ubuntu` for
Ubuntu-specific packages.
