Re: [RFC/PATCH 2/2] log: add "--no-show-signature" command line option

From: Mehul Jain <mehul.jain2029@gmail.com>
Date: May 27, 2016, 06:08 UTC
Message-ID: <CA+DCAeTnkTRnmscduio1-buKd2gNcaScYkLY7YfEQ+_Cm=Z28Q@mail.gmail.com>
In-Reply-To: <xmqqiny0wx8m.fsf@gitster.mtv.corp.google.com>

On Thu, May 26, 2016 at 10:52 PM, Junio C Hamano <gitster@pobox.com> wrote:

Show 37 quoted lines

> Jeff King <peff@peff.net> writes:
>
>> On Thu, May 26, 2016 at 06:36:47PM +0530, Mehul Jain wrote:
>>
>>> If "log.showsignature=true", then there is no way to override it using
>>> command line switch.
>>>
>>> Teach "git log" and "git show" about "--no-show-signature" command line
>>> option.
>>
>> I think this is teaching all of the revision machinery about it (which
>> is a good thing).
>
> I agree that the proposed commit log message should be updated to
> say so.
>
> Because we do not want .showsignature configuration to affect
> rev-list nor format-patch, and we will not make "--show-sig" the
> default for them either.  From that point of view, there is no
> reason for them to know about the "--no-show-signature" option.
>
> The only reason why teaching the "--no-show-signature" option to
> these commands is a good idea is because it would help people who
> create an alias with "--show-sig" in early part of the command line,
> e.g.
>
>         [alias] fp = format-patch --show-signature
>
> by allowing them to countermand with --no-show-signature, i.e.
>
>         $ git fp --no-show-signature ...
>
> If we are updating the log message in the final submission of this
> patch, we'd want it to be clear that the presence of this option is
> not an excuse to introduce .showsignature that affects rev-list
> later to make sure we do not have to waste our time rejecting such a
> patch in the future.

Currently, with the [patch 1/2], only git-show, git-log, git-whatchanged and git-reflog are able to learn about log.showsignature config variable. But commands which will learn about "--no-show-signature" with [patch 2/2] are notably a super-set of above mentioned commands. Introduction of this option should not give an impression that we might need log.showSignature for commands like git-format-patch etc, and it will definitely be a wise decision to convey the same in the commit message of this patch. I will do the necessary change.

Just out of curiosity, I was thinking that we might be able to teach "--no-show-signature" option only to git-show, git-log, git-whatchanged and git-reflog. To do this we can introduce a new member "no_show_signature" in struct rev_info, and use this variable further to modify the value of value of "rev.show_signature" after init_revision() is called. This way we can selectively decide which commands should learn about "--no-show-signature". This may be a bad idea because we will have two variables in rev_info, for option --[no]-show-signature. Any thoughts?

Thanks, Mehul