-----Original Message-----
From: ietf-dkim-bounces(_at_)mipassoc(_dot_)org [mailto:ietf-dkim-
bounces(_at_)mipassoc(_dot_)org] On Behalf Of Al Iverson
Sent: Friday, April 23, 2010 2:07 PM
To: ietf-dkim(_at_)mipassoc(_dot_)org
Subject: Re: [ietf-dkim] Why mailing lists should strip DKIM
signatures
On Fri, Apr 23, 2010 at 9:45 AM, Dave CROCKER <dhc(_at_)dcrocker(_dot_)net>
wrote:
On 4/23/2010 6:50 AM, MH Michael Hammer (5304) wrote:
If John is making some assertion of responsibility for his message
by
signing, what is the limit of his responsibility as the message
flows
through
the ecosystem? Where is the RFC that says his signature should be
stripped?
Most importantly, where is the specification that says a DKIM
signature
overrides The MailFrom address?
Not everything is codified in RFC or elsewhere. If John sends email to
my mailing list, and I emit that mail to the world, and it garners
complaints, it strikes me based on custom and history that I am the
responsible party. John would not be. Not directly, anyway.
But John made a private arrangement with Yahoo that if there was a
complaint about a mail and he DKIM signed it then Yahoo should send the
complaint to him as part of it's FBL offering. They did exactly what he
asked them to do.
If the list stripped his signature and someone modified what he
wrote
is this
a failure of DKIM or is it something else? What are we collectively
(and
individually) trying to achieve if we are signing the body and not
just
the
headers?
If a list already knows it should strip DKIM signatures, isn't also
likely that
the list will be able to sign?
No, because stripping the signature is currently easier than
generating a new one. Stripping the signature is just removing text.
Adding a new signature requires functionality not inherent to all MTAs
and MLMs.
The fact that it is easier does not make it correct - doesn't
necessarily make it incorrect either - that's in part what the
discussion is about. So if the list strips the signature and doesn't
sign itself then John's responsibility (which he asserted) is abrogated
with no acceptance of responsibility by the list owner. Is this really a
general behavior that we want to promote? I ask this in all seriousness.
We have no empirical data that the presence of a list signature AND
an
author
signature will produce the wrong results (for some definition of
wrong.)
Yeah, but clearly the author signature alone can cause what somebody
here thinks to be an imperfect result.
I tend to agree with him. I've been stripping DKIM signatures on my
own hosted mailing lists for that reason, and also so I can modify
content on the fly without the original signature failing.
But are you (people we can have a reasonable expectation that we can
somewhat trust to act responsibly) the rule or are you the exception?
I think I tend to agree with Steve. Notify all parties that assert
responsibility. That would include the author domain signer as well as
the list if they wish to accept responsibility for mail they emit.
Mike
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html