threads / discuss / 18030

git-push: creating branch on remote, refspec format

Subject: git-push: creating branch on remote, refspec format

## tl;dr

3 messages between Feb 26, 2009 and Feb 26, 2009.

replies: 2people: 3as markdown or json

Lasse Kliemann· Feb 26, 2009, 12:00 UTC · lore

http://www.kernel.org/pub/software/scm/git/docs/git-push.html reads:

| git push origin master:refs/heads/experimental
| 
|   Create the branch experimental in the origin repository by copying the 
|   current master branch. This form is only needed to create a new branch or 
|   tag in the remote repository when the local name and the remote name are 
|   different; otherwise, the ref name on its own will work.

It appears to me that the ref name on its own will also work to create a new branch, at least with local repositories:

$ (mkdir a && cd a && git init)
Initialized empty Git repository in /tmp/a/.git/
$ mkdir b && cd b && git init
Initialized empty Git repository in /tmp/b/.git/
$ pwd
/tmp/b
$ touch x && git add x && git commit 
[master (root-commit)]: created 87a90f5: "NA"
 0 files changed, 0 insertions(+), 0 deletions(-)
 create mode 100644 x

Now the push from '/tmp/b' to '/tmp/a':

$ git push /tmp/a master:foo
Counting objects: 3, done.
Writing objects: 100% (3/3), 209 bytes, done.
Total 3 (delta 0), reused 0 (delta 0)
Unpacking objects: 100% (3/3), done.
To /tmp/a
 * [new branch]      master -> foo

Am I missing something?

Michael J Gruber· Feb 26, 2009, 12:08 UTC · re: Lasse Kliemann · lore

Re: git-push: creating branch on remote, refspec format

Lasse Kliemann venit, vidit, dixit 26.02.2009 13:00:

Show 35 quoted lines

> http://www.kernel.org/pub/software/scm/git/docs/git-push.html reads:
> 
> | git push origin master:refs/heads/experimental
> | 
> |   Create the branch experimental in the origin repository by copying the 
> |   current master branch. This form is only needed to create a new branch or 
> |   tag in the remote repository when the local name and the remote name are 
> |   different; otherwise, the ref name on its own will work.
> 
> It appears to me that the ref name on its own will also work to create a new 
> branch, at least with local repositories:
> 
> $ (mkdir a && cd a && git init)
> Initialized empty Git repository in /tmp/a/.git/
> $ mkdir b && cd b && git init
> Initialized empty Git repository in /tmp/b/.git/
> $ pwd
> /tmp/b
> $ touch x && git add x && git commit 
> [master (root-commit)]: created 87a90f5: "NA"
>  0 files changed, 0 insertions(+), 0 deletions(-)
>  create mode 100644 x
> 
> Now the push from '/tmp/b' to '/tmp/a':
> 
> $ git push /tmp/a master:foo
> Counting objects: 3, done.
> Writing objects: 100% (3/3), 209 bytes, done.
> Total 3 (delta 0), reused 0 (delta 0)
> Unpacking objects: 100% (3/3), done.
> To /tmp/a
>  * [new branch]      master -> foo
> 
> Am I missing something?
>

You're using the very same form as the doc. "This form" in the doc refers to "repo refspec", i.e. something like "a b:c", which is what you use. The "refs/heads/" part is just there to disambiguate between possible different matches on the remote side (refs/heads/foo, refs/tags/foo, refs/remotes/bar/foo, ...).

Michael

Jeff King· Feb 26, 2009, 13:18 UTC · re: Michael J Gruber · lore

Re: git-push: creating branch on remote, refspec format

On Thu, Feb 26, 2009 at 01:08:29PM +0100, Michael J Gruber wrote:

Show 8 quoted lines

> > $ git push /tmp/a master:foo
> [...]
> 
> You're using the very same form as the doc. "This form" in the doc
> refers to "repo refspec", i.e. something like "a b:c", which is what you
> use. The "refs/heads/" part is just there to disambiguate between
> possible different matches on the remote side (refs/heads/foo,
> refs/tags/foo, refs/remotes/bar/foo, ...).

Actually, it used to be that case that git would complain unless "foo" was properly qualified; this was fixed in f8aae12 (push: allow unqualified dest refspecs to DWIM, 2008-04-23), which is in v1.5.5.2.

The doc in question was added around v1.5.3. So I think it is simply that it was never updated to match the new behavior. That paragraph could probably even just be deleted, since there is no syntactic difference between pushing to an existing branch and creating a new branch (alternatively, it can say explicitly that creating a new branch is the same as pushing to one).

-Peff