From: Jeff King <peff@peff.net>
Date: Tue, 11 Mar 2014 18:00:20 GMT
Subject: Re: [PATCH 4/7] commit: fix patch hunk editing with "commit -p -m"
Message-ID: <20140311180020.GA28154@sigill.intra.peff.net>
In-Reply-To: <CA+SSzV04KCGyTkFTtSx_9sYAZyh=UzHOPV6tf9JT82w_DsSLHg@mail.gmail.com>

On Tue, Mar 11, 2014 at 06:56:02PM +0100, Benoit Pierre wrote:

> According to the original commit, the change to GIT_EDITOR is only
> here for hooks:
> 
> commit 406400ce4f69e79b544dd3539a71b85d99331820
> Author: Paolo Bonzini <bonzini@gnu.org>
> Date:   Tue Feb 5 11:01:45 2008 +0100
> 
>     git-commit: set GIT_EDITOR=: if editor will not be launched
> 
>     This is a preparatory patch that provides a simple way for the future
>     prepare-commit-msg hook to discover if the editor will be launched.
> 
>     Signed-off-by: Junio C Hamano <gitster@pobox.com>
> 
> So there is really no reason to set it earlier, and not just in the
> hook subprocess environment.

Ah, you're right. I was thinking that our invocation of launch_editor
also respected it. It does, but we never get there at all because
use_editor is set to 0. So yeah, it really is just needed for the hook.

Your patch, even though it is a bigger change, keeps the setting to the
minimal area, which is cleaner and more maintainable in the long run.

-Peff