www-infrastructure-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Brett Porter <br...@apache.org>
Subject Re: @apache.org commit address requirement (Was: Git hosting is go)
Date Wed, 21 Dec 2011 10:12:13 GMT

On 21/12/2011, at 7:14 PM, Roy T. Fielding wrote:

> 
> I agree with everything except "ensure commits contain information about
> which ASF committer was involved (not for provenance, but for project members
> to know what is going on with a commit)".  I think that could be accomplished
> simply by using the pusher's Apache ID in the email notification that is sent
> to the commits list.
> 
> I wish that Git allowed appends to the commit log, but it doesn't, and there
> is significant value in not changing the patch identifiers (the hash). In
> fact, I'd argue that changing the commit hash is the primary con, because
> it is the one thing that truly ties the pushed content to the contributor's
> intent to contribute.
> 
> IOW, if I receive a pull request or changeset on a public list from a
> contributor that does not have a CLA on file, I want to preserve that
> changeset exactly as contributed so that the commit log will automatically
> refer to that third-party and the hash can be used to search for their
> contribution request.  I would only want to modify the changeset
> and provide a new log entry for those few cases where the existing
> information is not sufficient to satisfy my own CLA. [And, for that
> case, we should make it a good practice to include the original changeset
> identifier in the new log entry.]

Thanks - that's a much clearer justification than keeping the committer intact, which I hadn't
gleaned from the earlier conversation.

My limited experience with external contributions has been mixed between pull requests being
merged and rebased, and I got the impression rebasing was used more often than perhaps it
is. I also don't tend to maintain a repo that diverges from the "central" one, so I'm not
too concerned about the commit IDs on mine.

> 
> Regardless, the committer information is rarely shown by git -- you have
> to do something like
> 
>  git log --pretty=fuller
> 
> just to see it.  The commit hash is far more valuable information.

Right, but other tools are showing it. I was hoping for something like these:
https://github.com/brettporter/centrepoint/commit/6df5bf435e3a90ec4e036c19ad6cdfee2f3906fa
https://git-wip-us.apache.org/repos/asf?p=couchdb.git;a=commit;h=93b7f34bca748284c65eb2d471b33628a966aa0f

If it can be found by mail or some other web tool provided by infra by searching for the hash,
then I agree that's sufficient even if not ideal.

The only other mechanism I can think of to attach it to the history would be to force a merge
commit, but I don't like that option any better and it doesn't seem any more easily navigable.

Is there still some value in validating the committer emails against a project managed list
(the CONTRIBUTORS/pom.xml solution I mentioned earlier - which can automatically include anyone
with an ICLA and perhaps be optional per project), as a means of preventing accidents? The
nature of these being a little hidden in some git processes has bitten myself and others on
a number of occasions.

Cheers,
Brett

--
Brett Porter
brett@apache.org
http://brettporter.wordpress.com/
http://au.linkedin.com/in/brettporter
http://twitter.com/brettporter






Mime
View raw message