On Wed, 23 Dec 2020 at 14:04:45 +0000, Jelmer Vernooij wrote: >> If I understand the source correctly, this is because the gitconfig >> library it's using doesn't understand these settings. It might be a >> wishlist bug for the library, however lintian-brush could maybe call >> `git config user.email` instead and let the git binary take care of the >> parsing and fallback logic? > > Yep, that's correct. Right there is no dependency on C Git, so I'd > prefer to fix this in Dulwich (the library being used to read > ~/.gitconfig). That way, it also fixes this issue for other Git users.
Ack. The downside being of course that the gitconfig semantics is a
moving target and it might be non-negligible work to keep it up to date.
> The reason for the current implementation is to be consistent with the
> the way that dch (using DEBEMAIL) and debcommit (using
> GIT_COMMITTER_EMAIL/GIT_COMMITER_NAME) work, so it's least
> controversial.
Oh I see. Didn't realize debcommit was not honoring DEBEMAIL since it
uses C git which picks the correct user.email from the conditional
include :-)
> Would it make sense for e.g. debcommit to set GIT_COMMITTER_EMAIL to
> DEBEMAIL? Are there reasons why DEBEMAIL and GIT_COMMITTER_EMAIL
> should ever be different?
Not that I can think of, and same thing for GIT_AUTHOR_EMAIL and
GIT_{AUTHOR,COMMITTER}_NAME. Filed the suggestion at #977973.
Thanks,
--
Guilhem.
signature.asc
Description: PGP signature
