From: Jeff King <peff@peff.net>
Date: Thu, 25 Aug 2016 18:39:28 GMT
Subject: Re: git push origin BRANCHNAME question
Message-ID: <20160825183928.jm7pihc4kzwuej6y@sigill.intra.peff.net>
In-Reply-To: <xmqqr39cn3fk.fsf@gitster.mtv.corp.google.com>

On Thu, Aug 25, 2016 at 10:50:23AM -0700, Junio C Hamano wrote:

> Ed Greenberg <edg@greenberg.org> writes:
> 
> > I think I understand this from the git-push man page, but I want to
> > make sure:
> >
> > I have two branches, master and develop.
> >
> > If I am (accidentally) sitting on master, and issue 'git push origin
> > develop', does this properly push develop to remote develop, or does
> > it push master to remote develop (which seems to be bad, in the most
> > common use case.)  ?
> 
> You can find it out yourself quite easily, I would think.
> 
> $ git init src
> $ git init dst
> $ cd src
> $ git commit --allow-empty -m initial
> $ git checkout -b develop
> $ git commit --allow-empty -m second
> 
> $ git checkout master
> $ git push ../dst develop
> Counting objects: 3, done.
> Delta compression using up to 6 threads.
> Compressing objects: 100% (2/2), done.
> Writing objects: 100% (3/3), 226 bytes | 0 bytes/s, done.
> Total 3 (delta 1), reused 0 (delta 0)
> To ../dst
>  * [new branch]      develop -> develop

Yes, though I think the "why" is interesting here, too.

And the answer is that "develop" is a shortened refspec, whose full form
is more like "develop:develop". A name with no colon is always the local
source, and the implied destination is usually the same name on the
remote (though it can be changed via config). The third paragraph of
"<refspec>..." under OPTIONS in "git help push" covers this, though I do
think it is a little hard to follow.

-Peff