No. “?all” in an SPF record is a negative signal to many filters and a quick 
way to the spam folder. It also exposes the domain to abuse unconnected with 
DMARC.

If a sender intentionally relies on DKIM-only alignment, then that’s their 
decision. Making any recommendations as to what their SPF record should 
contain, other than being valid, is out of scope. Such recommendations are also 
prone to make operational assumptions. For example, assuming that the author 
domain is never going to be used in any other context, such as a 5321.From 
domain in a different message with a non-DMARC protected 5322.From domain. A 
specification tries to avoid such assumptions.

We also have ARC for cases where an intermediate MTA rewrites the 5321.From.

If we really need to spell out the potential risks of DKIM-only (or indeed 
SPF-only) alignment, then maybe a BCP document is a better place. It’s not like 
this is a widespread problem currently.

Ken.

From: dmarc <dmarc-boun...@ietf.org> On Behalf Of Douglas Foster
Sent: Friday 11 February 2022 08:14
To: IETF DMARC WG <dmarc@ietf.org>
Subject: [dmarc-ietf] (7.1?) DKIM-only authentication

I know that we took out the reference to default policy at my request, and I 
think it was in section 7.1.   But subsequent discussion helped me to 
understand objectives that were not clear to me in the previous text.   I think 
we need to re-insert something specific about domain owners that want DKIM-only 
authentication.   Proposed language:
“Some domain owners want DMARC authentication to use DKIM signatures only.   
This requires ensuring an SPF result other than PASS.  An SPF result of FAIL or 
SOFTFAIL is likely to produce unwanted rejects by non-DMARC evaluators.   An 
SPF result of NONE may be ineffective if an evaluator responds to NONE by 
applying a locally-defined default SPF policy that produces an unintended SPF 
PASS.   Domain owners who desired DKIM-only authentication are RECOMMENDED to 
publish a policy of “?ALL”, which ensures an SPF result of NEUTRAL, neither 
PASS nor FAIL.    Similarly, DMARC evaluators SHOULD treat SPF NONE as 
equivalent to NEUTRAL when the RFC5322.From domain has an applicable DMARC 
policy record.”

Doug Foster

_______________________________________________
dmarc mailing list
dmarc@ietf.org
https://www.ietf.org/mailman/listinfo/dmarc

Reply via email to