MX dot was (Re: [ietf-dkim] TXT wildcards SSP issues

[Steve Atkins]

On Jun 3, 2007, at 9:51 AM, Douglas Otis wrote:

> On Sun, 2007-06-03 at 08:58 -0700, Steve Atkins wrote:
>> (Still way OT, but it's more interesting than SSP, eh?)
>>
>>> IMO it's a really odd idea as it causes DNS root queries by  
>>> standards
>>> compliant MTAs and changes the semantics of MX (now all of a  
>>> sudden MX
>>> relates to sending mail).
>>
>> No, it shouldn't cause root queries.
>
> Paul Vixie was fairly convincing in explaining why this construct  
> caused
> significant trouble when used with SRV RRs for indicating "No  
> Service".
> The SRV draft even documents the meaning of the "." construct from
> inception.  Still, software must exclude interpreting this as a valid
> host.  When seen as a valid host too often, this construct can not be
> safely used, even when initially defined as "No Target."  For the MX
> record, this field is defined as a valid host.  Use of this construct
> would be expecting software to have the foresight to exclude ".".
>
> Yahoo gets a fair number of queries for "http://www.yahoo.com" A IN
> records which was fixed with a cname.  Ignorant software is  
> everywhere,
> why expect otherwise?

This is straying a long way from relevant to DKIM, so I'm not going to
try and explain the basics of DNS here.

Cheers,
   Steve