Is a complete false dilemma. Many protocols -- rfc822 included -- have multiple
outputs
and neither a) suffer from them nor b) require some artificial ranking system
of #1, #2,
ad nauseum. "Layer violation" is a shibolleth aimed at making DKIM into
something that
it never was: a single scaler return-value RPC call. The tags and values of the
DKIM
header and the signature-covered -- and uncovered -- parts of the message are
all
fair game to anybody who wants to use DKIM information.
This was by explicit intent at least on my part: I expected DKIM to be a tool
that
constrains the problem space of sending and receiving mail. Even within the
confines
of "take responsibility for" there is no reason to infer that means that there
is
a single scaler output of DKIM. In fact, the h= and l= tags completely negate
that
inference since I clearly do not take responsibility for the unsigned parts of
messages. And the only way to know that is to consider those tags as well.
To not do so on the pretense of "layering violation" is not only silly,
it is semantically at odds with the charter.
So please let's not hamstring DKIM with all kinds of adolescent navel gazing
about identity. The people who will use DKIM to their advantage are smart folks
and will figure the utility of the protocol out for themselves. What they
don't need is a school marm telling them which are the naughty and nice bits
of the protocol.
Mike
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html