https://bz.apache.org/SpamAssassin/show_bug.cgi?id=7139

RW <[email protected]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[email protected]

--- Comment #7 from RW <[email protected]> ---
Just to be clear, this is not a test for an IP address, it's a test for a helo
that is not RFC compliant. It should be 

  Received: from [127.0.0.1]  ... 

not

  Received: from 127.0.0.1  ...


If the brackets are there you have helo=!127.0.0.1!, not helo=127.0.0.1 and so
the test fails.

In the example given the fault shouldn't cause the rule to fire as the header
is in the internal network - the OP should check his settings if he's seeing
FPs. It may affect the deliverability of forwarded mail though and should be
fixed.

Spamassassin tests for RFC violations should IMO stand or fall by the
statistics. Personally I don't have a single instance of a legitimate mail
having helo=127. Unless someone has some evidence for broken headers with
127.0.0.0/8 being a disproportionate cause of FPs, I think this should be
reverted.


BTW __FSL_HELO_BARE_IP_2 should really be using X-Spam-Relays-Untrusted, since
there's no point in punishing faults in the trusted network when it can be
avoided. __FSL_HELO_BARE_IP_1 has to be left with X-Spam-Relays-External.

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to