Re: nnimap and searching articles by message-id

[Ted Zlatanov <tzz@lifelogs.com>]

On Mon, 21 Sep 2009 16:25:11 -0700 (PDT) janko hrasko <jhrasko1@yahoo.com> wrote: 

jh> --- On Sat, 9/19/09, Reiner Steib <reinersteib+gmane@imap.cc> wrote:

>> > It turns out that the IMAP server I am using 
>> 
>> Which IMAP server software is it?

jh> It is a part of Oracle Beehive collaboration suite.

>> 
>> > requires the message-id be stripped of the enclosing
>> angle brackets
>> > when searching for messages like so:
>> >
>> > UID SEARCH HEADER Message-Id "xxxx.yyyy@foobar.com"
>> >
>> > Not sure who is not compliant here: Gnus/nnimap or the
>> IMAP server?
>> 
>> Cc-ing the developer list.  Maybe someone there
>> knows.

jh> It would be good to know.  What do people think?  A quick scan through the RFC2822 - Internet Message Format reveals the following, but I am not sure if that is enough to convince anyone. 

jh>    Semantically, the angle bracket characters are not part of the
jh>    msg-id; the msg-id is what is contained between the two angle bracket
jh>    characters.

jh> Thanks for the patch.  It didn't work for what I needed (e.g. fetch
jh> the parent article) until I applied analogous changes to the other
jh> occurrence of search by message-id in nnimap.el:
jh> nnimap-request-article-part.

Can you and Reiner put together a patch?  Should be a simple one.

I think stripping off the angle brackets MAY cost some efficiency in
other IMAP servers (maybe they hash by the full message ID), but I don't
know.  The SEARCH reference in RFC 3501
(http://www.ietf.org/rfc/rfc3501.txt) says that search by any field is
by contained string, so as far as the standard is concerned stripping
off the angle brackets is harmless.  I would make it a user option
unless we're sure Courier, Dovecot, Exchange, and UW IMAP will work the
same way with and without the change.

Ted