Re: A Quickhack

On Sun, Oct 26, 2003 at 10:16:40AM -0500, Alan DeKok wrote:
> 
>   RFC 2821, Section 4.1.4, Paragraph 6 says:
... 
>   We don't want our proposals to violate RFC 2821.
> 


Well, isn't it the protocol defined by RFC2821, which allows
spam and fraud? Isn't that exactly what we want to stop?



>   In addition, there's no reason why there should be any correlation
> between IP address and host name.  Many people put SMTP servers behind
> NAT gateways, for example.
> 
>   All that is required for RMX is that the domain has a statement that
> this IP is permitted to use SMTP, and to claim to be from the domain.
> That's quite a different statement than "this hostname corresponds to
> this IP"


I'm afraid there is some confusion now. There are different opinions
about what to do.

If there is _no_ correlation between IP address and host name, what's
the point in verifying it?


To verify that this particular IP address is authorized to send
e-mail, we need a domain name anyway. If a HELO IPADDR is given, we
fail anyway. 

The NAT problem you mention is a little bit more complicated in my
eyes: The name must point to a valid DNS zone, otherwise DRIP will
fail as well. And if this zone exists, it will most likely contain a
A record. If this A record points to an RFC1918 address, this is normally 
wrong, because those addresses should never be published to the
Internet through DNS.



Maybe it will be easier to decide how we do it if we first decide, 
what exactly we want to do.

Hadmut