From: Carl Worth <cworth@cworth.org>
To: Thomas Schwinge <thomas@schwinge.name>, notmuch@notmuchmail.org
Subject: Re: [PATCH] Add a few tests for searching LWN emails.
In-Reply-To: <1296124309-20072-1-git-send-email-thomas@schwinge.name>
References: <1296124309-20072-1-git-send-email-thomas@schwinge.name>
User-Agent: Notmuch/0.5 (http://notmuchmail.org) Emacs/23.2.1
	(i486-pc-linux-gnu)
Date: Sat, 29 Jan 2011 05:59:38 +1000
Message-ID: <87vd18n6j9.fsf@yoom.home.cworth.org>
MIME-Version: 1.0
Content-Type: multipart/signed; boundary="=-=-=";
	micalg=pgp-sha1; protocol="application/pgp-signature"
Cc: lwn@lwn.net, Thomas Schwinge <thomas@schwinge.name>
Precedence: list

--=-=-=

On Thu, 27 Jan 2011 03:31:49 -0700, Thomas Schwinge <thomas@schwinge.name> wrote:
> These tests should pass -- but they currently don't.

Hi Thomas,

Thanks for sending these test cases. This is actually my favorite way to
receive bug reports. I really appreciate it!

> What we get from these emails, is an author named ``LWN.net'', and the
> ``Weekly Notification'' / ``Mailing Lists'' bits are stripped away.  I
> suspect this may be a misinterpretation in the notmuch address parser,
> related to the dot in the name.  I have not yet looked at the relevant
> code.

Yes, I believe this is related to the dot in the name. From my
recollection a name with an address requires quoting. So the header that
is currently formatted as:

	From: LWN.net Weekly Notification <lwn@lwn.net>

should instead be:

	From: "LWN.net Weekly Notification" <lwn@lwn.net>

I verified that adding this quoting fixes the tests[*]. I've CCed
lwn@lwn.net in case the kind editors there would like to add quoting
here.

I don't have any of the relevant RFCs in front of me now, so I don't
know exactly how an address with this missing quoting should be
parsed. But I recall having failures trying to send mail to an address
like this formatted without the quoting.

So I'm not sure what could reasonably be changed here in GMime or not. I
definitely do want to fix notmuch so that it indexes all of the text
here regardless if it's formatted in an RFC-compliant way or not.

-Carl

[*] Except for test VI which has a bug in that it searches for the word
"mailing" in the subject header, but no such word exists in the
message's subject header.

--=-=-=
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)

iD8DBQFNQyAq6JDdNq8qSWgRAo/XAJ9HPYSosTjl8QwYBqBsJmv6G2xB0wCffdm8
wK5/lhf1VR9BKv0bviGpuM8=
=Hxun
-----END PGP SIGNATURE-----
--=-=-=--