Re: dgit and false attributions (was: rescue_1.106_source.changes ACCEPTED into unstable)
On Sat, Dec 27, 2025 at 10:59:45AM +0100, Cyril Brulebois wrote:
> Are we really OK as a project (Debian) or as a team (installer team)
> to have commits that are created on behalf of others, without their
> knowledge or consent?
Hi. I'm just a lurker here, and I'm not even a user of dgit, but yes,
I think it is definitely correct to create commits on behalf of others
in *some* cases.
For example, whenever some package is hosted in salsa for the first
time, somebody usually takes all releases from snapshot.debian.org and
creates a "fake" git history from it, using gbp import-dsc or some
similar tool.
Those tools try to make a history which resembles as much as possible
the history that the real uploaders would have created during the life
of the package, so I see as a natural thing that a commit which matches
a given upload is attributed to the person who authored the upload.
If your complain is that the commiter should be different than the
author, well, I would say that it's a legitimate complain, even if I'm
not sure that having the committer to be the same as the author is
necessarily wrong.
I suggest you contact dgit developers for that, maybe they can provide
the algorithm (and the rationale) used by dgit to decide author/committer.
Thanks.
Reply to: