Re: [PATCH] checkpatch: improve handling of email comments

From: Joe Perches
Date: Fri Oct 30 2020 - 08:22:14 EST


On Fri, 2020-10-30 at 12:58 +0100, Lukas Bulwahn wrote:
> On Fri, 30 Oct 2020, Joe Perches wrote:
> > On Fri, 2020-10-30 at 14:37 +0530, Dwaipayan Ray wrote:
> > > checkpatch has limited support for parsing email comments. It only
> > > support single name comments or single after address comments.
> > > Whereas, RFC 5322 specifies that comments can be inserted in
> > > between any tokens of the email fields.
> > >
> > > Improve comment parsing mechanism in checkpatch.
> > >
> > > What is handled now:
> > >
> > > - Multiple name/address comments
> > > - Comments anywhere in between name/address
> > > - Nested comments like (John (Doe))
> > >
> > > A brief analysis of checkpatch output on v5.0..v5.7 showed that
> > > after these modifications, the number of BAD_SIGN_OFF warnings
> > > came down from 2944 to 1424, and FROM_SIGN_OFF_MISMATCH came
> > > down from 2366 to 2330.
> > >
> > > So, a total of 1556 false positives were resolved in total.
> >
> > A mere reduction in messages emitted isn't necessarily good.
> >
>
> Agree. That is why I also went through the list of those warnings.

So sending me a copy of that list shouldn't be a burden.

> > Please send me privately a complete list of these nominally
> > false positive messages that are no longer emitted.
> >
> > I believe one of the relatively common incorrect messages is
> > for the cc: <stable@xxxxxxxxxxxxxxx> where a version number is
> > continued on the same line after a #.
> >
> > CC: stable@xxxxxxxxxxxxxxx # for versions x.y.z and above
> >
>
> That was one, another common pattern was just quotes put inconsistently at
> different places.

Which to me is more an indication that a message
_should_ be emitted as many email clients do not like
to copy/paste incorrectly formatted email addresses
(ie: Missing necessary quotes when the name contains
characters like .) and that's a common way to cc a
reply to a possible commit message of an email.

Perhaps as well the .mailmap mechanism may not cope
with these differently formatted email addresses.