Re: [PATCH 0/7] Make "$remote/$branch" work with unconventional refspecs

From: Junio C Hamano <gitster@pobox.com>
Date: May 5, 2013, 22:36 UTC
Message-ID: <7vvc6xt5ov.fsf@alter.siamese.dyndns.org>
In-Reply-To: <CALKQrgf6NcT2tEGMTczxR2WspOi4NjrN_kxmKN-QyE2Py3iSaQ@mail.gmail.com>

Johan Herland <johan@herland.net> writes:

> This would not allow the user to use the relevant $remote_name for $nick,
> which I argue might be the more natural name for the user to use, since
> it's the same name that is used for otherwise interacting with the remote.

That is where we differ.

The thing is, when you name a local ref (be it "refs/heads/master" or "refs/remotes/origin/master") with a short-hand, you are still dealing with a refname, not interacting with the remote at all.

Taking notice of remote.$nick.fetch mappings only to complicate the refname resolution logic is absolutely unacceptable, at least to somebody who comes from the "we are interacting with refs, not with remotes" school, like me.