Re: [RFC/PATCH] verify-tag: add --check-name flag

From: Jeff King <peff@peff.net>
Date: Jun 7, 2016, 22:29 UTC
Message-ID: <20160607222908.GA25631@sigill.intra.peff.net>
In-Reply-To: <xmqqk2i0od1f.fsf@gitster.mtv.corp.google.com>

On Tue, Jun 07, 2016 at 03:21:48PM -0700, Junio C Hamano wrote:

Show 9 quoted lines

> Jeff King <peff@peff.net> writes:
> 
> > If you are suggesting that you can do the whole thing today by parsing
> > the tag object yourself, then sure, I agree. I thought the point of the
> > exercise was to make that less painful for the callers.
> 
> Yes, and I somehow thought everybody agreed that --show-tag-name was
> striking the balance at about the right level for ease-of-use and
> simplicity?

No, I think "--format" would be much better, unless you want to add a separate "--show-tagger-ident" when somebody wants to do a check between the tagger's ident and the key uid.

But either way, I think the whole "do a rev-parse first" thing raises the question of what object identifiers "git tag" would accept. We would presumably expect:

  git tag --show-tag-name v1.0

to work. And I think in your world-view, so would:

  git tag --show-tag-name $(git rev-parse v1.0)

How about:

  git tag --show-tag-name refs/tags/v1.0

And what about:

  git tag --show-tag-name refs/remotes/foo/v1.0

or even:

  git tag --show-tag-name foo/v1.0

when refs/remotes/foo/v1.0 exists?

The rule right now is generally that "git tag" takes actual tag names. Plumbing like "verify-tag" takes arbitrary get_sha1() expressions, but you're expected to qualify or resolve your refnames before you get there, to avoid weird situations. This "tag --show-tag-name" seems to sit in the middle of plumbing and porcelain (for that matter, I am not sure that it should belong to git-tag at all, as it is really about scripting).

-Peff