From: Junio C Hamano <gitster@pobox.com>
To: Patrick Steinhardt <ps@pks.im>
Cc: git@vger.kernel.org
Subject: Re: [PATCH v3 4/5] builtin/hash-object: fix uninitialized hash
 function
In-Reply-To: <ZkLpVSVv1XPz-D8v@tanuki> (Patrick Steinhardt's message of "Tue,
	14 May 2024 06:32:21 +0200")
References: <cover.1715582857.git.ps@pks.im>
	<20240513224127.2042052-1-gitster@pobox.com>
	<20240513224127.2042052-5-gitster@pobox.com>
	<xmqq7cfxcnlc.fsf@gitster.g> <ZkLpVSVv1XPz-D8v@tanuki>
Date: Tue, 14 May 2024 08:55:45 -0700
Message-ID: <xmqq8r0c5qwu.fsf@gitster.g>
User-Agent: Gnus/5.13 (Gnus v5.13)
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain

Patrick Steinhardt <ps@pks.im> writes:

> I'd prefer the latter -- always use SHA-1. As you say, it's easier to
> understand and doesn't create implicit mechanisms that we'll have to
> maintain going forward. Also, users didn't have a desire yet to pick a
> different algorithm than SHA-1, which probably also comes from the fact
> that SHA-256 repositories are still scarce.
> ...
> Eventually, we should then add a new option `--object-hash=` to
> git-hash-object(1) and other commands that may run outside of a Git
> repository to let the user pick their desired hash.

If we were planning to allow using something other than SHA-1 in the
future, then I do not think I agree with your argument.  We can use
the same GIT_DEFAULT_HASH mechanism to specify what hash to use without
later adding an extra option just fine.