www-infrastructure-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jukka Zitting <jukka.zitt...@gmail.com>
Subject Re: @apache.org commit address requirement (Was: Git hosting is go)
Date Thu, 15 Dec 2011 00:23:25 GMT

On Wed, Dec 14, 2011 at 11:24 PM, Paul Davis
<paul.joseph.davis@gmail.com> wrote:
> If I read your case right, then the confusion would be when 8de5e722ea
> already exists in the repository.

Right. The current setup doesn't really give us an easy and
straightforward record of who pushed each commit to the canonical
repository. Anyway, as said, this information can be deduced from the
current records, so it's not really a fundamental issue. It just
seemed to me like a possible reason for the restrictions on the %ce

> I don't think this is the same thing. The ref-updates.log contains who
> updated what refs during a push. This is separate from who made the
> commit. Ie, the point of checking %ce in the commit is to check that a
> commit was entered by a project committer but this importantly not the
> same as who's making the push. Ie, this allows a committer to push
> updates that include commits by other committers.

The problem I see with this is that it prevents commits made by
non-committers (anyone can commit with Git) to be pushed without these
commits being rebased or otherwise mangled.

IMHO we shouldn't be putting restrictions on who can commit (in the
"git commit" sense) to a repository. Instead the restriction (and
related audit trails) should be on who can make those commits a part
of the canonical repository on git-wip-us (i.e. "git push"). That's
more in line with the distributed nature of Git.

> I'm not sure I follow. The %ae field should always be the person that
> wrote the commit. The %ce is who made the commit.

Right. With Git the one who made the commit could well be someone
who's not an official committer. (Yes, the traditional terminology is
breaking down here...)


Jukka Zitting

View raw message