On Sun, 8 Feb 2004, Michael Quinion wrote:

> One of my subscribers missed confirmation messages from my LISTSERV
> because his ISP was blocking all incoming traffic that didn't contain
> a message ID. Their reasoning was that it was an effective way to
> trap messages containing the MyDoom worm, which don't contain an ID
> line. Unfortunately, LISTSERV housekeeping messages (and digests,
> too, I believe) also don't have message IDs. A rather snappy message
> came back to me from their support people, asking what the rationale
> was for LISTSERV messages not to confirm to the recommendations of
> RFC 2822 & RFC 2119. Would anyone care to comment?

The Message-id is allowed by RFC to be added by the first SMTP host that
handles the message, if it is not already present. It is not the
responsibility of LISTSERV necessarily, but whatever MTA you are using to
distribute the SMTP jobs if you want every message to have a Message-ID.

RFC 2822 makes the Message-ID a "should" not a "must" (see section 3.6.4).

RFC 2119 defines "should" rather well.

Blocking messages because they lack a Message-ID is a lame way to block
spam and viruses. I have about 2900 spam messages (including 30 virus
messages without payload) in my archive (for training Bayes databases) and
none lack the Message-ID. About 4% of my "good message"  archives lack a
Message-ID.



--
Douglas Palmer              |
SystemsManager, USDC-EDNY   |
225 Cadman Plaza East       |   Internet: [log in to unmask]
Brooklyn, New York 11201    |   CCNA,CCNP