Re: git push mirror mode

From: Andreas Ericsson <ae@op5.se>
Date: Nov 8, 2007, 13:14 UTC
Message-ID: <47330BA4.6030101@op5.se>
In-Reply-To: <20071108124435.GH9736@shadowen.org>

Andy Whitcroft wrote:

Show 23 quoted lines

> On Thu, Nov 08, 2007 at 12:19:18PM +0000, Johannes Schindelin wrote:
>> Hi,
>>
>> On Thu, 8 Nov 2007, Andy Whitcroft wrote:
>>
>>> Ok, sometime back Junio sent out a proof-of-concept change to
>>> send-pack allowing a mirror mode.
>> You added/left his sign-off, but did not attribute the patches to him.  
>> Why?
> 
> I believe I left his signed off by from the original (first) patch, and
> added mine to indicate that what I had modified was also unecombered.
> The second patch is only signed off by me as I am the author.  In my
> world (admittedly a kernel hacker) the first Signed-off-by: indicates the
> primary authorship of that patch and the [apw@...] part tries to clarify
> the changes I made therein.
> 
> No intentional stripping of credit was intended, and I believe that the
> attribution as written states Junio is the originator of this patch.
> However that is the way I would read the meanings of these lines, if git
> has different rules or you think there is a clearer way of stating this
> I am happy to change it, and resend it so attributed.
>

Barring any errors in my understanding of the matter, here's how it works for git.

git separates author from committer, so code attribution is done with author, and "I verified this is sane" is done by committer. Those two usually only ever differ when the user tells git commit that the author was someone else than him/her self, or when rewriting history with git rebase or similar. git am also maintains authorship (using the From: line in emails), but sets $committer to the person running it, so when you apply patches sent by email from someone else you get the code attribution right by default.

The Signed-off-by line is, in git, used as "I touched the code here and agree that it may be included in the mothership repo and all future releases" (the spirit of that sentence is also in Documentation/SubmittingPatches).

We also have Acked-by (as does the kernel, no? I think we inherited it from there) to mean something along the lines of "I vote we include this", but not always based on technical merit (ie, patches can have many acks without having ever been tested).

Suggested-by, Tested-by and Reported-by are used less often, not always written in dash-form, but hopefully always self-explanatory ;-)

-- 
Andreas Ericsson                   andreas.ericsson@op5.se
OP5 AB                             www.op5.se
Tel: +46 8-230225                  Fax: +46 8-230231