Re: [PATCH 1/8] git-merge-file --ours, --theirs

From: Junio C Hamano <gitster@pobox.com>
Date: Nov 26, 2009, 07:05 UTC
Message-ID: <7vvdgxbwav.fsf@alter.siamese.dyndns.org>
In-Reply-To: <20091126153726.6117@nanako3.lavabit.com>

Nanako Shiraishi <nanako3@lavabit.com> writes:

Show 9 quoted lines

> Quoting Junio C Hamano <gitster@pobox.com>
>
>> Except for parse-optification, this one is more or less a verbatim copy of
>> my patch, and I think I probably deserve an in-body "From: " line for this
>> [PATCH 1/8], [PATCH 6/8] and [PATCH 8/8] to take the full authorship of
>> them.
>
> Could you give an guideline to decide when to take authorship and when to
> give it to others?  The above seems somewhat arbitrary to me.

It might seem that way to you, as you do not write much C, but I am reasonably sure Avery understands after having worked on the series.

Imagine that Avery were an area expert (the subsystem maintainer) on "git merge" and downwards, and somebody who did not know that "merge" has already been rewritten in C, nor some parts of the system have been rewritten to use parse-options, submitted a patch series for review and Avery is helping to polish it up [*1*].

As the subsystem maintainer, Avery looks at the patches, updates parts of the code that is based on obsolete infrastructure, adds lacking tests and documentation as necessary, and forwards the tested result upwards for inclusion. How would the messages from Avery to me look?

Patches that were majorly reworked should be attributed to Avery, and obviously new patches that are added for missing tests should be, too. For example, changes made to git-merge.sh by the original submitter must be discarded and redone from scratch to builtin-merge.c, and if you look at the changes, it would be quite obvious that the original patch wouldn't have served as anything more than giving the specification.

But the ones with minor updates should retain the original authorship. It unfortunately is not black-and-white, though.

In any case, where does Avery's credit go? Is there a point in helping to polish others' patches?

It is recoded on the Signed-off-by line. When somebody passes a patch from somebody else, an S-o-b is added for DCO purposes, but it also leaves the "patch trail"---these people looked at the patch, spent effort to make sure it is suitable for inclusion by reviewing, polishing, and enhancing. A subsystem maintainer, or anybody who helps to polish others contribution, may not necessarily have his name as the "author" of the patch, and if the patch forwarding is done via e-mail, his name won't be on the "committer" line either. But the contribution is still noted and appreciated, and the hint to pay attention to is by counting non-author S-o-b and Tested-by lines in the commit messages.

cf. http://lwn.net/SubscriberLink/363456/50efdecf49af77ba/ check the last table.

[Footnote]

*1* That somebody happens to be me from 18 months ago, but the important point here is that the person is not Avery as the subsystem maintainer (in other words, it is immaterial that it happens to be the same person as the toplevel maintainer).