From: Philip Oakley <philipoakley@iee.email>
Date: Sat, 24 Aug 2019 13:59:01 GMT
Subject: Re: [PATCH v3 13/13] format-patch: learn --infer-cover-subject option
Message-ID: <1eccefe4-8002-4664-3bc2-8332fb7bd697@iee.email>
In-Reply-To: <20190824080328.GA9726@archbookpro.localdomain>

On 24/08/2019 09:03, Denton Liu wrote:
> On Fri, Aug 23, 2019 at 01:18:44PM -0700, Junio C Hamano wrote:
>> Philip Oakley <philipoakley@iee.email> writes:
>>
>>> Perhaps the `--infer-cover-subject` the config option needs to be
>>> multi-valued to include:
>>>       "subject" (always expect short first lines) or
>>>       "message" (always the long paragraph description, still use
>>> ***Subject Here***),
>>>       with the "true" being used when expecting both as previously
>>> described.
> Good idea, I like this a lot!
>
>> The idea to have three choices feels that this is getting better,
>> but I notice that the choice is no longer about "subject".
>>
>> I've always felt that the name of this option is way suboptimal.
>> One reason is because the option only says it is about the subject
>> of the cover (letter), and the verb "infer" conveys almost no
>> information---especially it does not say anything about what affects
>> the inference (hint: the branch description value gets used, in a
>> single hardcoded ways right now, but now with the patch we have a
>> choice to control how it gets used).
> Perhaps something like
> --cover-subject-from-description={true,auto,false}?

maybe --cover-letter-from-description={true,auto,subject,message,false}? 
to cover most eventualities (i.e. letter rather than subject).

I haven't looked at what happens on Windows (CRLF usage?) for multi-line 
descriptions. The common assumption is LF in repo, with attributes etc, 
but the branch description is a bit free format in terms of guidance ;-)

Philip