From: Junio C Hamano <gitster@pobox.com>
Date: Mon, 13 Aug 2012 07:26:11 GMT
Subject: Re: [PATCH 0/4] Re: cherry-pick and 'log --no-walk' and ordering
Message-ID: <7vd32vfe24.fsf@alter.siamese.dyndns.org>
In-Reply-To: <7vhas7fefs.fsf@alter.siamese.dyndns.org>

Junio C Hamano <gitster@pobox.com> writes:

> Would anything break if we take your patch, but without two
> possibilities to revs->no_walk option (i.e. we never sort under
> no_walk)?

By the way, by "would anything break", I do not just mean if our
existing tests trigger failures from "test_expect_success"; I
suspect some do assume the sorting behaviour.  I am wondering if the
sorting makes sense in the real users; in other words, if the
failing tests, if any, are expecting sensible and useful behaviour.

After all, the sorting by the commit timestamp is made solely to
optimize the limit_list() which wants to traverse commits ancestry
near the tip of the history, and sorting by the commit timestamp is
done because it is usually a good and quick approximation for
topological sorting.