At 20.33 +0200 98-08-17, Ned Freed wrote:
Received: fields, like Resent- fields are only added to primary message
headers, which existing email signature mechanisms do not cover.
Does that mean that existing signature standards will not
guarantee that the subject or the CC fields have not been
tampered with?
It depends on which fields you are talking about. If you are talking about
fields at the outermost level, then yes, existing signature standards have no
way to sign them. However, there is nothing to prevent you from using an extra
wrapper around the actual message if you want to secure these fields. You can
even wrap up the entire SMTP transaction if you like and tunnel it through
inside a message.
Is that acceptable?
It certainly appears to be.
And how about the date,
that could have legal significance in some cases, why
should it not be protected?
The date claimed by the message originator has little if any significance in
most cases. What would matter would be, say, a date added by a timestamping
service. But in such cases encapsulation of the message along with timestamp
information is essential. And existing signature protocols handle this just
fine modulo a somewhat more specialized timestamp container specification --
the tricky part is the timestamp service and the liability issues it faces.
Also the From and and the
In-Reply-To fields can be quite significant. The intrinsic
meaning of a message with the text "Yes, I agree" can be
quite different depending on the value of the In-Reply-To
header.
Again, the From: address claimed by the originator is of little signifcance.
And similarly, for the in-reply-to to have tamper-proof meaning imposes
requirements not only on the in-reply-to field but also on what it refers to.
You can build systems that secure this sort of thing using existing protocols,
but it remains to be seen whether anybody thinks it important enough to
actually do so.
Jacob, all of these points and literally hundreds of others have been discussed
ad nauseum in the development of PEM, MSP, security multiparts, MOSS, S/MIME,
PGP-MIME, and OpenPGP. We're talking 10+ years of standards work and heaven
only knows how many messages -- hundreds of thousands at a minimum. I suggest
you review the (vast) record in this area. This is neither the time nor the
place to discuss these matters.
Ned