[Exim] Re: spamassassin but not system wide

Página superior
Eliminar este mensaje
Responder a este mensaje
Autor: Richard, WhidbeyNet NOC
Fecha:  
A: exim-users
Asunto: [Exim] Re: spamassassin but not system wide
-----Original Message-----
Richard, WhidbeyNet NOC said:
>> A different technique we looked into was modifying SA-Exim to scan
>> messages using local_scan() during the connection phase. Using that
>> method, "spamd" would return the message to Exim with a modified
>> header, and never reject the message. Then later on, we'd scan for

the
>> "X-Spam" header, depending on a users preference.
>>
>> When scanning at the connection phase, it's difficult to tell who has
>> filtering turned on, because at that stage Exim only has an address
>> (not a final local user). That means most messages need to be scanned


>> (any local domain). That, combined with the fact that we want to
>> preserve Exims integrity, and the need for SA-Exim to be heavily
>> modified, turned us away from local_scan().
>
>What kind of modifications does SA-Exim needs? If all that you want is
>to scan the message, add the X-Spam headers and then >accept it, it can


>easily be configured to do this.
>
>The easiest way to scan every message and only mark them would be
>using:
>SAEximRunCond: 1
>SAEximRejCond: 0
>
>What more did you want?
>--
>Patrice Fournier


That's a good point to make, and we could find a way to make that work.
However, we need the ability to use a users SA prefs, and it's easiest
to do that when run at delivery. We also don't want to run SA for users
who don't want it, and want to avoid relying on a header that could
(although unlikely) be forged. It mostly comes down to personal
preference, and how much mail you need scanned I suppose.

We also want to run aditional filtering programs, that have no SA-Exim
equivalent, which is one reason we're looking at a direct-spool access
solution. Maybe Exim will incoporate ability that in the future? To pass
messages to other programs for delivery decisions, like Procmail does.

Rich
richs@???