spf-discuss
[Top] [All Lists]

RE: Re[2]: OT: education needed

2004-11-30 17:44:19
On Tue, 30 Nov 2004, David Woodhouse wrote:

On Tue, 2004-11-30 at 09:14 +0000, Richard Bang wrote:
Our MTA goes one further. It will only resolve to a domain if the admin has
specifically bound the IP to a domain. Otherwise it always fails.

The _only_ working localpart at the IPv4 or IPv6 literal domains of my
mail servers is postmaster(_at_)(_dot_)

There's no requirement that anything else should work, and certainly no
requirement that a given localpart at any or all of the virtual domains
hosted on any given box shall map to the same destination mailbox or
forwarding address as each other or to the same destination as the same
localpart at an IP literal domain.


I would actually go one step further and say that you don't really have to 
accept mail for ip literals like @[123.4.5.6].  If there is a requirement to 
accept such mail I don't think I have seen it.

I believe that if you choose to send mail FROM such a domain, it should be 
considered valid syntax by the receiver.  For example, if you don't have a 
domain name at all, or the sending server doesn't know its own domain name, it 
can still send out using an address literal.  (If you can't afford a lawyer, 
one will be provided to you by the court, but he might smell funny.)

I believe that if you send mail OUT using such a suffix, then you ought to be 
able to accept return of that mail, and be able to accept postmaster(_at_)[x] 
if 
you have used [x] as an ip literal in outgoing mail in the last 90 days.  I 
think that if you don't use ip literals in outgoing mail, there's no need to 
accept them.

Really the use of an address literal in a return address or From: address 
should be deprecated.  I think that if any receiver chooses not to accept mail 
from an address literal, that should be their choice.  (They can also choose 
not to accept mail from Nigeria or China.)  It's not against the rules to send 
it, but it's also not against the rules to refuse to accept it.

It would be quite silly in my opinion to accept mail from 
user(_at_)123(_dot_)4(_dot_)5(_dot_)6 where 
the [square brackets] aren't present - that is clearly not valid.  I think 
it's also silly to accept mail from user(_at_)[10(_dot_)0(_dot_)0(_dot_)1] 
(rfc1918 is not internet) 
or user(_at_)[127(_dot_)0(_dot_)0(_dot_)1] (same) or any bracketed IP that 
doesn't match the actual 
sending client's IP.  Sure, a valid message might come along, but the sender 
of such a message is pretty much always doing something wrong and has an 
alternative for getting the message through more correctly.

that's my 2 cents on the topic. :)

--
Greg Connor
gconnor(_at_)nekodojo(_dot_)org

Everyone says that having power is a great responsibility.  This is a lot
of bunk.  Responsibility is when someone can blame you if something goes
wrong.  When you have power you are surrounded by people whose job it is
to take the blame for your mistakes.  If they're smart, that is. 
                -- Cerebus, "On Governing"


<Prev in Thread] Current Thread [Next in Thread>