Re: [Evolution] Evolution generating invalid message-ids

On Sun, 2002-07-28 at 04:34, Not Zed wrote:
So where's the bit in the email that relates to the subject of the

In my email or in the one from the balsa list?

Jörg was suggesting that it's not a good idea (actually, he said it's
"wrong") to have the rhs of the message-id header be "localhost" or
"dirk" because that's increasing the chance of another message having
the same message-id. The page by jzw suggests a way of doing it which
comes very close to guaranteeing uniqueness.

Jörg furthermore suggested that it's not even Evo's business to generate
Message-Id headers.  The RFC's say it should be generated by Evolution. 
Nevertheless, there might be situations where it would be better if
Evolution didn't generate a Message-ID header, as witnessed by the
problem reported in the balsa thread, where a user's (undoubtedly,
non-standards-compliant and misconfigured) SMTP server just wouldn't
accept messages with Message-ID headers set.  

There are good reasons why you don't want to not generate Message-IDs
(eg, it breaks threading client-side, it's not sanctioned by the RFCs).
I'm just saying that generating Message-IDs may get you into trouble
(well, not you, but users might): they might not be able to send through
their SMTP server or their messages might not get past some spam
filters, either because those servers are misconfigured or because the
RHS of the Message-ID doesn't pass DNS checks (as they would if the RHS
was "dirk" or "localhost").  I know, Jeff fixed it so that it does a DNS
lookup on the hostname. However, if that fails, it falls back to
"localhost".  I suspect that that means that anyone on a dialup
connection working offline gets their Message-IDs set to localhost


On Sat, 2002-07-27 at 10:13, Richard Zach wrote:
Just FYI:  Balsa had a similar wasn't accepting
messages for delivery if they have a Message-ID.  In the ensuing
discussion, the possibility of adding a config option "Do not generate
Message-ID" was considered. It wasn't implemented. The thread starts

Jamie Zawinski wrote something for DRUMS on message-id's:
(I know this is an Internet Draft and there's no reason whatsoever to
do anything recommended there--other than that Jamie Zawinski is a
pretty bright guy.)

From what Matt Curtin and he are saying, it's clear that they take
"globally unique" in RFC 822 to mean actually unique, not just unique
for the generating host.  They suggest to use the host/domain part of
the user's address as a fallback if you can't get the FQDN of the host.
They also say that "use of an unqualified hostname for the domain part 
of the Message-ID header would be foolish, and should never be done." 
By extension, I take it, using "localhost" or "localhost.localdomain" 
for the rhs of the message-id would not be advisable.


evolution maillist  -  evolution ximian com

Richard Zach ......
Assistant  Professor,   Department  of  Philosophy
University of Calgary, Calgary, AB T2N 1N4, Canada

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]