[ietf-dkim] 1365, with a question about the "we never send mail"

[Steve Atkins]

On Oct 14, 2006, at 9:57 AM, Scott Kitterman wrote:

> On Saturday 14 October 2006 12:38, Stephen Farrell wrote:
>
>
>> Having said that, our job for now is to figure out whether to include
>> this requirement or not for SSP, and your mail isn't entirely  
>> clear as
>> to whether you think SSP needs to support that requirement.
>
> I'm ambivalent.  I think it [that a domain claims to send no mail]  
> is a fact
> that would significantly aid receivers in evaluating received  
> mail.  I don't
> know that it needs to be defined here.

It's a pretty rare case. I think the main
reason it keeps coming up is that it's the only case in which a sender
domain can make a statement that will work reliably (in either SPF
or SSP), so is used as the "But look! It can work in this case!" poster
child to support the need for that sort of protocol, rather than because
many people actually need the functionality (either as senders or
recipients).

> I think there is some harm in every e-mail identity related protocol
> re-inventing how to express 'sends no mail'.  I think it would be  
> better to
> do it once.  Given that there is one method that is standardized at  
> least at
> the experimental level, in theory I think it's better not to try  
> and reinvent
> the wheel here.  In practice, I understand that would open a rather  
> large can
> of worms.
>
> Just to be clear, I was suggesting using a TXT record with the sting
> literal 'v=spf1 -all' in it and not suggesting trying to drag the  
> entire
> protocol along.

If you do that, you're dragging the whole SPF protocol along, whether
you mean to or not, for recipients at least (unless I'm  
misunderstanding,
which is perfectly possible).

Cheers,
   Steve