On Jun 16, 2009, at 10:51 AM, SM wrote:

> At 08:00 16-06-2009, Cullen Jennings wrote:
>> My focus is mostly on what this errata means to implementations in  
>> the field. That would be implementations of both DKIM the narrow  
>> signing protocol and implementations that use the information DKIM  
>> provides. I think the document should be clear on
>
> The proposed update does not affect the DKIM verification process.   
> It can affect the DKIM signing process and the post-verification  
> process.  I'll use your message as an example.  The DKIM-Signature  
> header for your message contains "d=cisco.com;   
> [email protected];"  According to RFC 4871, the domain is taken  
> from the "i=" tag and in the absence of that tag, we fallback to the  
> "d=" tag.  For this message, we would use "cisco.com" for the  
> assessment.  I'm taking a narrow view of the assessment.
>
> It was pointed out that some DKIM signers use a DKIM-Signature  
> header such as "d=example.com; [email protected];".   
> According to RFC 4871, we would use ["core5.example.com"] as the  
> domain for the assessment.  With the proposed update, the "i=" tag  
> and the fallback behavior is ignored.  We use "example.com" for the  
> assessment.

Correct, but this change would reduce DKIM protections!

The intent of the i= parameter is to permit finer grain header and  
intra-domain resolution.  This resolution ability is lost when always  
defaulting to the use of d=.  The use of the d= parameter does not  
differentiate between a mailing list of [email protected] and  
that of a user of example.com,  such as:

  Sender: [email protected]
  From: [email protected]
  dkim-signature: [email protected]; [email protected];

Neither reputation OR annotations are better handled when limited to  
JUST the d= parameter.

The i= parameter, when not included, would be "@<d=value>".  This is  
the only time that d= should be used!


> Some DKIM signers may have to change the way they were signing   
> messages if they are passing domain related information through the   
> "i=" tag instead of the "d=" tag.  The DKIM post-verification  
> process also has to be modified to pass the domain from the "d=" tag  
> instead of the one from the "i=" tag.

This does not work unless separate sub-domain and selectors are setup  
to contain separate keys.  This also means that users of example.com  
messages will have a valid signature when signed by a subdomain of  
example.com.  As such, this type of change will necessitate the use of  
sub-domains and multiple signatures, which the i= conventions would  
have avoided.

>> 1) what is the interoperability problem. Can someone succinctly
>
> See the above example.

The above example creates problems.  It does not solve any.  The i=  
value remains optional, and does default to the d= value anyway!

>> summarize this? When I read the document, I get that there is a  
>> problem but it less clear what it is or why some implementation  
>> would end up doing something that did not work with other  
>> implementations.
>
> Some people got creative. :-)

Some people are attempting to mitigate intra-domain spoofing.  A REAL  
problem!   DKIM is better than path registration, so don't emasculate  
this difference!

>> 2) what needs to be changed in implementations to fix this? Again,  
>> can anyone succinctly summarize this. I do not think an implementor  
>> that
>
> See the above example.
>
>> did not follow the list could easily read the current draft and  
>> figure out if there code was OK or not and what changes where  
>> needed to their code to make it OK
>
> To keep it simple I'd say "use the d= tag only".

This will not allow domains that might have an intra-domain problem  
from establishing finer grain resolutions, or to clarify whether a  
message was sent on-behalf-of their list, or on-behalf-of some user  
within their domain.

>> I'm not really sure what you mean by a reputation intent and non- 
>> repuation intent or when a signer or verifier would want one or the  
>> other.
>
> reputation intent - help us blacklist, whitelist you, etc.
>
> non-reputation intent - the DKIM signer placed some obfuscated  
> information in the "i=" and it was probably not meant to be used for  
> reputation.

Yes, the i= value CAN be used for intra-domain reputation.  This would  
occur only when the domain is otherwise good, but when there might be  
problems with some accounts having fun with replay abuse.

-Doug



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

Reply via email to