Re: [PATCH] Add tests for git format-patch --to and format.to config option

From: Junio C Hamano <gitster@pobox.com>
Date: Mar 7, 2010, 03:42 UTC
Message-ID: <7vaaukdch2.fsf@alter.siamese.dyndns.org>
In-Reply-To: <4B92EDFA.1000602@gmail.com>

Stephen Boyd <bebarino@gmail.com> writes:

Show 10 quoted lines

> On 03/05/2010 06:21 PM, Junio C Hamano wrote:
>> This reveals that --to does not follow the usual rule to override
>> corresponding configuration.  Is that really what we want?  IOW, when the
>> command line says scipient, shouldn't we stop sending to recipient that
>> comes from the configuration?  How else would a user override this?
>>
>> So I guess the topic wasn't ready for 'next' yet, after all.
>
> The same applies to the fomat.headers and format.cc config options. How
> is this different?

Not different. Perhaps we should fix them now you noticed they share the same problem?

An obvious alternative is to keep format.to not get overriden by --to as the original patch did; that would at least make the handling between config and option consistent inside the command, but at the same time, it means format-patch behaves differently from everything else in git.

I don't have strong preference either way myself, and while I know "fixing" the ones you listed would affect existing users, I have this suspicion that it wouldn't be too big a problem. After all, format-patch does not _send_ mails; these MUA-like "features" doesn't belong to the program, and nobody should be relying on them heavily in the first place.