On Thu, Oct 09, 2025 at 12:13:48AM +0300, Adrian Bunk wrote: > On Wed, Oct 08, 2025 at 10:24:36PM +0200, Santiago Vila wrote: > >... > > (If this was not a team-maintained package, I would force-push > > as needed to match the archive). > > Commits removed by force-push stay in all forks and clones, this can > create a mess with more than one committer or merge requests enabled.
By "not a team-maintained package" I was actually thinking about my own repositories, where there is only one commiter, and I have not merge requests enabled. For the general case, yes, one should be careful not to disrupt the work by others. Michael: The current repository has a commit saying "Acknowledge NMU" to refer to 2.7.2-1.1, but none of the current commits matches exactly version 2.7.2-1.1. I think that's not ok and should be fixed, so I'm going to add a single commit to match 2.7.2-1.1, then I will add the proper tag, then I'll add the NMU from Adrian which has just arrived to unstable, also with tags. Then we will be in sync with the archive. [ After that, I would recommend not add commits like "new upstream" unless such new upstream is actually close in time to being actually uploaded ]. Fortunately debichem repositories usually contain only debian/* files, so undoing a "new upstream" which did never actually happen is not particularly troublesome. Thanks.
