From: Junio C Hamano <gitster@pobox.com>
Date: Fri, 29 Jul 2016 17:06:44 GMT
Subject: Re: Small trivial annoyance with the nice new builtin "git am"
Message-ID: <xmqqbn1gpe57.fsf@gitster.mtv.corp.google.com>
In-Reply-To: <20160729002902.GD9646@sigill.intra.peff.net>

Jeff King <peff@peff.net> writes:

> On Thu, Jul 28, 2016 at 04:47:17PM -0700, Junio C Hamano wrote:
>
>> Also makes me wonder if "git cherry-pick A..B" shares the same
>> breakage.
>
> Probably.

It seems that "cherry-pick A..B" leads to sequencer.c::run_git_commit()
that uses run_command_v_opt() to drive "git commit", so we are safe.

> I guess we want something like:
>
> +void reset_ident_date(void)
> +{
> +	strbuf_reset(&git_default_date);
> +}
> +
>
> and then to sprinkle calls liberally through builtin-ified programs when
> they move from one unit of work to the next.

ident_default_date() is currently the only one that sets this to be
cached, and that is to be used only when there is no user-specified
date.

When I saw the suggestion first time, I was worried if this had
interaction with things like GIT_COMMITTER_DATE environment (because
I didn't have easy access to the source) but it is not the case, so
the change looks very sensible.