Re: Registration model, 2821bis-06

Hi Arnt,
At 04:11 02-04-2008, Arnt Gulbrandsen wrote:
> How many people would be happy if 5.1 were to say this?
>
>  - if there is no MX record, but there is an A, then the sending
>    host MUST synthesise an MX
>  - if there is no MX record and no A, but there is an AAAA, then
>    the sending host MAY synthesise an MX
>  - even if a client doesn't use AAAA records for synthesising MX
>    records, it MAY use them for delivering mail once an MX has
>    been synthesised.
You may have noticed that draft-09 carefully sidesteps the IPv6 
question by mentioning an obviousness test from an operational 
viewpoint.  The above sounds more like a decision to be made during 
implementation.  The MAY makes it more difficult, from the server 
end, to determine why a message isn't being delivered as we have to 
know whether the client is designed to synthesize the AAAA RR as an 
MX in addition to the usual failure points.
> Connectivity to/from AAAA-only mail servers will suck terribly for 
> many decades to come, but no 2821bis wordsmithing can avoid that. 
> Only an arrangement with a dual-stack relay can improve 
> connectivity, and that practically requires an MX RR or more.
That's a reason for having the MX RR.

This discussion has covered both sides of the (implicit MX) argument 
from historical, email architecture, implementation and operational 
viewpoints.  Whichever option is accepted will be viewed as a change 
from RFC 2821.  I would be happy if the issue could be resolved one 
way or another.
Regards,
-sm