From: Paolo Bonzini <pbonzini@redhat.com>
Date: Wed, 19 Feb 2020 20:53:37 GMT
Subject: Re: [PATCH 3/4] am: support --show-current-patch=raw as a synonym for--show-current-patch
Message-ID: <217229b8-3a72-55fb-71c6-8ba8ae3ceb0b@redhat.com>
In-Reply-To: <xmqqmu9ee3hc.fsf@gitster-ct.c.googlers.com>

On 19/02/20 21:17, Junio C Hamano wrote:
>>> +       if (resume->mode == RESUME_SHOW_PATCH && new_value != resume->sub_mode)
>>> +               return error(_("--show-current-patch=%s is incompatible with "
>>> +                              "--show-current-patch=%s"),
>>> +                            arg, valid_modes[resume->sub_mode]);
>>
>> So, this allows --show-current-patch=<foo> to be specified multiple
>> times but only as long as <foo> is the same each time, and errors out
>> otherwise. That's rather harsh and makes it difficult for someone to
>> override a value specified earlier on the command line (say, coming
>> from a Git alias). The typical way this is handled is "last wins"
>> rather than making it an error.
> 
> Yup, the last one wins is something I would have expected.  And if
> we follow that (which is the usual pattern), I suspect that we won't
> even need the first two steps of this series?

We would need them anyway, in order to add a callback to the "command
mode" option --show-current-patch.

The fact that --show-current-patch is a command mode option is also why
I decided against "last one wins".  I think it would be counterintuitive
that

	git am --abort --show-current-patch

fails, but

	git am --show-current-patch=diff --show-current-patch=raw

succeeds.

Another possibility is to have separate options --show-current-message
(for .git/rebase-apply/NNNN) and --show-current-diff (for
.git/rebase-apply/patch), possibly deprecating --show-current-patch.
That would have naturally rejected a command line like

	git am --show-current-message --show-current-diff

(and this one _would_ have removed the need for the first two patches in
the series).  However, the long common prefix would have prevented using
an abbreviated option such as "--show", so I went instead for the
optional string argument.

I realize now that I should have placed all this in the commit message,
sorry about that.

Paolo