Switch to Exim4 for your MTA. It won't Append 2 files to the same
directory if it can see they both route to you.

Mark.

On 21 Nov 2007, at 08:50, Bjørn T Johansen <btj@havleik.no> wrote:

On Wed, 21 Nov 2007 09:05:14 +0100 (CET) Steffen Kaiser <skdovecot@smail.inf.fh-bonn-rhein-sieg.de> wrote:

...

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On Tue, 20 Nov 2007, Bjørn T Johansen wrote:

...

When I receive a mail that is addressed to my email address and
also with a cc to a mailinglist I subscribe to, I get two mails, which I guess is expected... But I have just moved
to Dovecot from Cyrus, and Cyrus seems to know that this is the same mail, so only one of those two mail is
stored on the imap srv...

I guess you mean: http://cyrusimap.web.cmu.edu/twiki/bin/view/Cyrus/DuplicateDeliveryExplained

Yes, this is what I mean..... :)

...

?

As far as I know there is no such thing in Dovecot's deliver.
However, deliver is easy to understand and probably you find someone to jump
in and make a patch.

If you do not use Dovecot deliver, check your local MDA.

Yes, I use Dovecot deliver....

...

If you mean that Cyrus stores two identical mails only once
physically, e.g. when spooling for two users, then no, because Dovecot is not
designed to only support virtual users, but with real users two users cannot
easily share the same file.

Nope, this is not what I mean....

BTJ

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On Wed, 21 Nov 2007, Mark Adams wrote:

Switch to Exim4 for your MTA. It won't Append 2 files to the same directory if it can see they both route to you.

Well, I expect any MTA to deliver a mail just once, if two recipients evaluate to the same mailbox. But to detect two deliveries in two separate mails is another problem and, because MessageIDs are not as globally unique as they should, there is no fail-proof method in order to detect them.

Bye,

Steffen Kaiser -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux)

iQEVAwUBR0Uqli9SORjhbDpvAQIh+gf/WrbDwgMbTXS+OGjYiqGP+/eFxblxPaWu 7NMjLf3CQc0iyeqtlf6GrG+vgK4goqNPcPLN7AQ7PUH6YSptAHZT1PWX29nLLX6M QZ4sXXN009JJa07j+Y0Rl4lUq6vveFfw+Vepv6haUCRx62tDLQtbyKioKA5JumyH 8goBms2G37/Mmosf2igrXiUe0CI6ls6Jnov76753mY8sRNwUST+XY/qgN63/+27U GThoxoDkBBlLEBqdAnyQaMJGXRu9cZ4MJbCfJUYWqkRa3Q8OlCFrT5qvzCt8/kdx BQXjbaEsbnxdD5nMxkEXJB6MIP9lhR764v2D2ErJA+QQKorjNcYpnA== =4zhM -----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On Thu, 22 Nov 2007, Nikolay Shopik wrote:

...

evaluate to the same mailbox. But to detect two deliveries in two separate mails is another problem and, because MessageIDs are not as globally unique as they should, there is no fail-proof method in order to detect them.

Well in case for example dovecot mail list when you reply by pressing reply all, it generate same message-id for these two messages. Anyway I'm agree with you, we can't be sure for all such cases.

The message ID is to be created by the sender, hence, is should be the same for both mails. (If the ID is added by a MTA, it's a SPAM indicator ;-) )

But although message IDs are required to be "globally unique" by RFC, it's a piece of data created by plenty of individual MUAs, bulk mailers, selfmade scripts etc.pp. Each implement a different strategy to acquire the "global uniqueness" (e.g. by putting together local hostname, mail domain, current time). But no such method is really fail proof. And at least - it is data any user can fake.

Bye,

Steffen Kaiser -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux)

iQEVAwUBR0VHSy9SORjhbDpvAQLergf/UlD9DgtcWm5ia7tqdFxEJTg3jNJRWzYS nByUPyw1x4xvNV2ix5wNqrPgaGU17neMSWNNTJnIDwUH7AXq/uio6SmWUzAKiBbN S4ZEj0zfjjJcsYDHQT7QDUKzfFY3npTseaMGShGgYNyU+L+bUfRWo04ReYobme65 hYKfIkrIvU6lrKTLTEcJQMD2+5POyFx6o3zaEcbetH1dZ9QbBGQESbvFVNtwa7XW ce+81vVODr9Q62w7rycWgKlNwDjpQrZVpl0Yx7YkLk7rtmFfkSRA8DAYrUZJuTFM uhEKKyz4BlNdahlA8tHzgvzLhioY8AmsNbR4ZHfRKtFH3jem+elpwQ== =Y1bo -----END PGP SIGNATURE-----