Re: [exim] Changing Email Identity

Magnus Holmgren wrote:

> On Wednesday 11 October 2006 20:28, Magnus Holmgren took the opportunity to
> say:
>
>>On Wednesday 11 October 2006 19:59, gascione took the opportunity to say:
>>
>>>We use exim4 servers as front end antivirus, spam checking,
>>>sender/receiver, stuff like that before the mail is passed off to our
>>>commercial email application. If a valid message is delivered the headers
>>>screw up some functionality on the mail server side because the last hop
>>>of the mail is our gateways which is always valid and therefore messes up
>>>the SPAM and filtering system on the mail server side.
>>
>>What systems are those? If they're decent (like SpamAssassin), it should be
>>possible to configure them to interpret the headers correctly.
>>
>>
>>>Is there a way to get exim to deliver the message to the mail server as
>>>if it was the original delivering mail server so the original header
>>>information remains at the top of the header when it reaches our mail
>>>server. Iin other words, can exim act like the original sending mail
>>>server when it passes the message along to our internal mail server.
>>
>>If you must, set received_header_text to nothing.
>
>
> Wait a minute ... That won't help. If your commercial mail application looks
> at the headers at all, and get it wrong, it's the line added by the internal
> server you have to get rid of.
>
> If that's not possible, then I think you have to start messing on the TCP/IP
> level and make it look like the connection from the Exim server actually
> comes from the remote host. It's theoretically possible, but ...
>
>

Might this serve just as well - and with less risk:

1) AFAIK, Exim has, or *can extract* the information that was in the 'Received:'
header just prior to the one it normally adds.

2) "Received:" headers are ordinarily 'stacked' - latest on top.

3) IF THEN Exim is told to save, then construct and add-back an extra
'Received:' header that is basically a duplicate of what it had on-arrival, that
header can be at the top of the stack.

Will that 'satisfy' the next level of servers w/o losing the 'real' traceback info?

... and, if so, might that be better than stripping headers?

Bill