Quoting from Matthias Andree's mail on Wed, Sep 04, 2002 at 09:44:49PM +0200:
Multidrop with wildcard can of course yield this 550 condition even
without configuration error (at least on Postfix), and some MTAs use
indeed 550 for content filter rejects as well as "no such user", so it
might be needed to add some parsing the reply message here, I'm thinking
of RFC-1893 (those codes used in Delivery Status Notifications), and/or
regexp matching (what is spam, what is not), for example.
Other possibility is to have an antispam per smtp command. For
example, 550 to MAIL FROM: should be in antispam, 550 to RCPT TO:
should not be, 550 to DATA should be.
Other possibility is to not do any antispam checks on RCPT TO:
command. I think the current antispam spam option is only meant for
response to the MAIL FROM: and DATA commands.
Here is a patch which does this last possibility. This patch now also
lets handle_smtp_report() send an RSET first as this is not being used
to handle response to RCPT TO: commands.
Sunil Shetye.
fetchmail-5.9.14-disablespamrcpt.patch
Description: Text document