ietf-dkim
[Top] [All Lists]

[ietf-dkim] Assessors

2009-06-19 03:40:54
Murray,

I just wanted to touch base on something I missed in your last post. A 
statement you made that now totally contradicts mthe model I thought 
you were talking about.  Maybe this discussion can help mold a better 
understanding for DKIM integrators.

EXECUTE will now have to create the proper function prototype stack
with the variable list parameters in order to satisfy specific
ASSESSOR APIs with different functional prototypes.

No.  Each assessor will feed a message to the DKIM API and get
back the result and "d=".

I thought I understood your model. I am not so sure now with the above 
statement which made be read more carefully what the ERRATA is saying 
and I simply can not comprehend it.

This somewhat conflicts with the layered DKIM model I had in mind 
which was DKIM WG chartered direction of splitting DKIM into a base 
spec to be integrated with higher layer policy, reputation and other 
"assessment" technology as there invented, materialized and applies 
to DKIM-BASE.

I thought "Assessors" were things like policy and reputations 
protocols and others yet to be invented, that assess the verification 
result of DKIM signature.

The thought the only required feed for DKIM-BASE was the 5322 payload:

   5322 -> DKIM-BASE -> status, d= -> DKIM-ASSESSOR -> PASS/FAIL

and the ERRATA discussions was about what DKIM verification result and 
additional data is needed for the next Assessor layer which ultimately 
provided final pass/fail (accept/reject) result.

Your statement seems to describe something different.

Does the Assessors do verification, policy, reputation and anything 
else with a final result of status and the d= tag?

If so, I was completely off base on what I thought this ERRATA was 
above. I can see better why you described your API as so, but IMV this 
is a prime example why should keep the API metaphor out of the errata 
and get this better cleared up.  It appears your idea of an API 
perfectly shows where there would be conflicts with a more straight 
forward understanding of DKIM layered framework.

    SMTP feeds DKIM-BASE, the result and additional data feeds a
    higher layer, which I thought were "assessors" but apparently
    not.

Assessor is really not the best term for this because it seems to 
applies it does everything, including the verification process and I 
guess it does, right?

Sorry I didn't understand your API, but it is very odd. This ERRATA is 
really complex and incomprehensible and in my opinion, it does seem to 
somewhere alter implementations API libraries I explored, like the 
ALT-N DKIM API library.  This API used the basic idea of a three layer 
approach:

    SMTP FEED (5322 Message)
      --> DKIM-BASE verification
           --> Optional DKIM-POLICY Evaluation Hook (one only)

I thought we were talking a simple change to better support future 
verification result analyzers, evaluators, policy, reputation, etc:

    SMTP FEED (5322 Message)
      --> DKIM-BASE verification
           --> Optional DKIM-ASSESSOR Hooks (multiple possible)

Whatever.

--

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

<Prev in Thread] Current Thread [Next in Thread>