[ietf-dkim] Discussion of Consensus check: Domain Existence Check

[Steve Atkins]

On Jun 9, 2008, at 8:55 PM, Scott Kitterman wrote:

> On Monday 09 June 2008 20:11, Frank Ellermann wrote:
>> John Levine wrote:
>>>> change the SHOULD [check MX & A/AAAA] to a MAY. With
>>>> that change, I'm happy with the text John proposes.
>>>
>>> OK by me.
>>
>> Hm, it is a border case, why not just say "MUST either
>> check [2821bis] or as minimal approximation [nxdomain]".
>>
>> If you really like MAY better I think "MAY instead" is
>> clearer than only "MUST [nxdomain], and MAY [2821bis]".
>>
>> If 2821bis is only a MAY you can't rename the result to
>> "nomailfqdn", it has to stay as is (= "nxdomain"), and
>> we can close that part of #1579, "not more applicable".
>>
> I'd like to suggest that since the how seems to be controversial we  
> punt on
> that and just say the ADSP only applies to domains that exist.  I  
> think we
> perhaps we can leave how to determine this as an implementation  
> detail.  The
> last 1% of agreement/interoperability isn't worth it.

That's an interoperability problem. If senders don't know what
recipients may check, then the ADSP check is ill-defined.

It does not really matter what the check is, as long as it's
well defined.

Cheers,
   Steve