On Jan 19, 2007, at 3:57 AM, Stephen Farrell wrote:


Thanks Eric,

Eric Allman wrote:
> I have submitted draft-ietf-dkim-base-08 for publication. This has some
> significant changes requested by the IESG.  Probably the most
> significant is the addition of section 5.4.1, "Recommended Signature
> Content".  The changes are summarized in Appendix F.1.

6. Verifier Actions     
                                
 Since a signer MAY remove or revoke a public key at any time, it is    
 recommended that verification occur in a timely manner. In many                
 configurations, the most timely place is during acceptance by the      
 border MTA or shortly thereafter. [In particular, deferring                    
 verification until the message is accessed by the end user is                  
 discouraged.]                  

This precaution should be removed!! When the public key record is revoked due to abuse, then not immediately verifying a message offers what might be valuable time. It is also likely the _only_ sure means to protect recipients from spoofing is through annotations applied by the MUA based upon recipient trusted email-addresses. Do not forget about the EAI changes and the vast range of resulting look-alike attacks made possible.

While trust might be established between the MDA and MUA, DKIM does not provide a means for making that determination. Without the ability of the MUA to also validate signatures, then any annotations applied based upon assumed trust in the MDA and some headers that may or may not be added imperils the recipient. Don't make the mistake that sender policy offers a reasonable solution either. Annotation based upon added headers or assumed application of sender policy reduces security. Rather than discouraging deferred verification, discourage the rapid removal of keys! There is an expiry that can be applied in lieu of key removal which can be compared against when the message was delivered. There is no valid reason to rapidly remove valid keys.

One of the benefits of DKIM is that annotation does not need to rely upon headers that may or may not be added by the MDA, which of course may or may not be also signed (and for how long?).

-Doug
_______________________________________________
NOTE WELL: This list operates according to http://mipassoc.org/dkim/ietf-list-rules.html

Reply via email to