From: Junio C Hamano <gitster@pobox.com>
Date: Fri, 10 Feb 2012 05:48:44 GMT
Subject: Re: What's cooking in git.git (Jan 2012, #08; Tue, 31)
Message-ID: <7v8vkbs0g3.fsf@alter.siamese.dyndns.org>
In-Reply-To: <CACsJy8B_nsZnSarbRaBoxviK1xa_PVP0-DbnMiP9HkwzrztfJg@mail.gmail.com>

Nguyen Thai Ngoc Duy <pclouds@gmail.com> writes:

> No objections. Does it fix a failed test? If not, should we add a test
> to record this side effect fix?

No, it was just "I tried this and it did not work" reported elsewhere, and
it is not v1.7.9 regression. Having seen a failure by a real user, having
a test would be a good idea to protect the fix from regressing.

I actually wish that the commit 9e58504 (clone: --branch=<branch> always
means refs/heads/<branch>, 2012-01-16) wasn't taken hostage to the earlier
changes that add new feature (addition of --single-branch and delaying of
the cloning before checking the remote HEAD); which would have been an
easier sell without violating the usual "no new features to maintenance
track".  And the thing is, I do not find this grave enough an issue that
deserves a separate implementation of a fix to be queued to maintenance
tracks.

> --mirror implies --bare in cmd_clone() if I read it correctly.

Yeah, what was I thinking... feeling stupid.

Thanks.