Re: [PATCH] Switch receive.denyCurrentBranch to "refuse"

From: Junio C Hamano <gitster@pobox.com>
Date: Feb 3, 2009, 04:30 UTC
Message-ID: <7vvdrsdtvr.fsf@gitster.siamese.dyndns.org>
In-Reply-To: <20090202124148.GB8325@sigio.peff.net>

Jeff King <peff@peff.net> writes:

> In both cases, you have a sequence of commands that does one thing with
> one git version, and something else with another git version. The only
> difference is whether your sequence includes git init.

I vaguely remember arguing against different behaviour between a new repository and an existing one in the past on a different topic myself. I am not married to the idea of effectively flipping the default to "refuse" in a new repository early, and do not mind dropping the "git init" change at all. I do not like the inconsistency myself.

The only reason why I did that "git init" patch was because I just thought that it might be a good way to help new people sooner, who will start using git after 1.6.2 gets released but before 1.7.0 flips the default for everybody, while explaining people older than 1.6.2 what is happening in the warning/error message during the transition period. I suspect it could be argued that with an extra line that says "the default will change to 'refuse' in 1.7.0 for all repositories, but we are making the change early for newly created repositories to help new people", the main idea of the patch may be salvageable, but I do not deeply care either way.

By the way, I just realized one thing.

When we flip the default to "refuse" in 1.7.0 for everybody, we will need the explanation and instruction on how to get a non-default behaviour and how to squelch the message when we "refuse by defaut", just like my first patch did when we "warn by default". It is entirely possible some people simply skip 1.6.2 and directly jump to 1.7.0, and while we cannot help them avoid the surprise caused by the change in behaviour, we cannot be silent in such a situation.