Re: Should `@` be really a valid git tag name?

From: Jeff King <peff@peff.net>
Date: Sep 17, 2021, 21:47 UTC
Message-ID: <YUUM5VqTEtjhnfjj@coredump.intra.peff.net>
In-Reply-To: <xmqq4kaihqjj.fsf@gitster.g>

On Fri, Sep 17, 2021 at 01:53:52PM -0700, Junio C Hamano wrote:

Show 19 quoted lines

> Carlo Arenas <carenas@gmail.com> writes:
> 
> > On Fri, Sep 17, 2021 at 7:12 AM <gitmailinglist.bentolor@xoxy.net> wrote:
> >>
> >> A SO commenter pointed out, that git-check-ref-format forbids @ and
> >> maybe I should report this as a potential bug. Is it?
> >
> > a reference that is named "@" only is invalid, but refs/tags/@ is not.
> 
> ;-)  
> 
> "git check-ref-format master ; echo $?" would show that any single
> level name is "forbidden", so probably the SO commenter (whatever
> that is) was confused---it is not about @ at all.
> 
> In any case, a tag whose name is @ may be another source of
> confusion in the modern world, after we added @ as a synonym to
> HEAD.  I do not know, for example, offhand which between the HEAD or
> that tag "git show @" would choose.  It makes sense to avoid it.

In the past when we've had confusing names (like refs/heads/HEAD), we continue to allow them at the plumbing level (to retain backwards compatibility), but flag them at the porcelain level to prevent users shooting themselves in the foot. This seems like a good candidate for that (for both git-branch and git-tag).

-Peff