[RFCI-Discuss] acceptable filtering of RFC2142 mailboxes?

[Derek J. Balling]

On Jul 3, 2006, at 2:20 PM, James Ralston wrote:
> Therefore, we are thinking of implementing a restriction that a
> message must either have *all* RFC2142 recipients, or not have *any*
> RFC2142 recipients.  If your message includes both RFC2142 and
> non-RFC2142 recipients, you will receive the following error after
> sending the "." to close the "DATA" command:
>
> 550 5.7.1 mixing RFC2142 recipients and non-RFC2142 recipients not  
> permitted; please resend to recipients individually
>
> My read of the rfc-ignorant abuse/postmaster guidelines is that
> filtering on this basis would *not* be grounds for being listed.

I would say that the better way to handle this is to return a 421  
code instead of 550 on that. As someone else pointed out, it's  
entirely possible that the following scenario occurs:

<SENDER> sends mail to <USER at YOURDOMAIN>.
<SENDER> finds their message spam-blocked in some fashion.
<USER> is highly annoyed, etc., etc.
<SENDER> sends mail to <POSTMASTER at YOURDOMAIN>, cc'ing  
<USER at YOURDOMAIN> so that <USER> knows that <SENDER> is on it.

This is a legitimate CC, and shouldn't, I think, be penalized.

By 421'ing the "opposite" (e.g., if your first RCPT is a 2142, then  
421'ing all non-2142 addresses, or vice versa), you force the sending  
MTA to resend the message a second time, but (presuming that you  
accepted all the RCPTs you didn't 421) it'll all happen  
transparently, without causing an unnecessary break in legitimate mail.

Does that make sense?

Cheers,
D