Re: [PATCH v2 1/1] branch: introduce --show-current display option

From: Jeff King <peff@peff.net>
Date: Oct 11, 2018, 20:53 UTC
Message-ID: <20181011205323.GB11618@sigill.intra.peff.net>
In-Reply-To: <20181011203518.GA2385@rigel>

On Thu, Oct 11, 2018 at 09:35:28PM +0100, Rafael Ascensão wrote:

Show 5 quoted lines

> On Thu, Oct 11, 2018 at 01:51:36PM -0400, Jeff King wrote:
> > Yeah, I agree.
> 
> Not sure which parts you meant, so I'll assume you didn't agree
> with me.

Correct. ;)

I like your general idea, but I agree with Daniel that it introduces an inconsistency in the interface.

Show 7 quoted lines

> I doesn't seem far fetched to ask for an overview of my current branch,
> feature1, feature2 and all hotfixes with something like:
> 
>   $ git branch --verbose --list HEAD feature1 feature2 hotfix-*;
> 
> The 'what's my current branch' could be just a particular case of this
> form.

Right, I like that part. It's just that putting "HEAD" there already has a meaning: it would find refs/heads/HEAD.

Now I'll grant that's a bad name for a branch (and the source of other confusions, and I think perhaps even something a few commands actively discourage these days).

Show 8 quoted lines

> My defense to treat HEAD specially comes in the form that from the user
> perspective, HEAD is already being resolved to a commit when HEAD is
> detached (Showing the detached at <hash> message.)
> 
> Is there a strong reason to *not* "resolve" HEAD when it is attached?
> Would it be that bad to have some DWIM behaviour here? After all, as
> HEAD is an invalid name for a branch, nothing would ever match it
> anyways.

I don't think this is about resolving HEAD, or showing it. It's about the fact that arguments to "branch" are currently always branch-names, not full refs.

-Peff