Using the From: address specified in the body

[Mauro Carvalho Chehab]

Em 17-03-2011 19:20, Peter Maydell escreveu:
> On 17 March 2011 23:06, Guilherme Salgado <guilherme.salgado at linaro.org> wrote:
>> On Wed, 2011-03-16 at 13:50 -0300, Mauro Carvalho Chehab wrote:
>>> Em 16-03-2011 11:03, Guilherme Salgado escreveu:
>>>> I see that some emails with patches have a From: field in the body[1],
>>>> and I'm wondering if there's any reason for not using that (when
>>>> available, of course) as the patch submitter.  Well, now that I think of
>>>> it, one could argue that the submitter is whoever sent the email, but
>>>> maybe it would be useful to have a 'author' field on Patch so that we
>>>> can properly represent cases where submitter != author?
> 
>>> For example, I've seem a few patches that have email references. It would
>>> not be impossible to see a "From: " or "Author: " in the middle of such
>>> references.
>>
>> That's a good point, but I think we could avoid that if we were
>> conservative and just used a 'From:' when it's at the beginning of a
>> line and that line is before the beginning of the diff.

Being at the beginning of a line is not enough. Patches that are forwarded
in general follows rule of having the From: as the first line of the
email body.

> Presumably what you actually want to do is accept From: lines in the
> body in exactly the cases where 'git am' accepts them, since that's
> the patch-email syntax where this case actually apperas in practice...
> git am is more restrictive than just "at beginning of the line and
> before the diff".
> 
> Here's an example where the From: in the body and the email
> From: don't match (not even the same person), because the patch was
> written by Christophe but (re)transmitted to the list by me as part
> of a larger patchset:
> http://patchwork.ozlabs.org/patch/83254/

Yeah, that patch is OK: First body line is a "From:" line, and the first
SOB is from the patch author.

> Guilherme: presumably Linaro will want to make a distinction
> between "Linaro wrote this" and "a Linaro person is doing
> upstream submaintainer type work with othre peoples' contributions"
> in our patch counting metrics? (if so, there's your test case :-))
> 
> For completeness, should we support the git am "Subject: can
> be at the start of the body" syntax too?

I think that, if such support is added on patchwork (both from: and subject:
replacements), the better would be to output them as a patchwork-specific
meta-data at the emails, like:

patchwork_subject:
patchwork_from:

This allow people that use some sort of script (like me) to decide how
they want to handle it.

Mauro.