[Top] [All Lists]

RE: Re[2]: IETF S/MIME and Algorithm Independence

1997-05-02 14:18:43
On Friday, May 02, 1997 9:09 AM, Paul E. Hoffman 
Paul, the current structure mandates the use of RSA key management. The
should not prohibit the use of other techniques such a Diffie-Hellman.  It
possible to have RSA as the "must implement" key management algorithm, but
standardized structure should not prohibit the use of other techniques.

And it doesn't. In fact, with the SMIMECapabilities, it makes it easy for
an agent that is using some other kind of key management to announce that
to the world on every signed and/or encrypted message it sends out. This is
described in the beginning of Section 2.5.2 of the drafts.

I think that Russ' (and Bill's original) point was that the PKCS #7
architecture has some things built into it that make it hard to use key
management other than RSA.  For instance to identify the certificate
used to encrypt the session key for a recipient, an
"IssuerAndSerialNumber" is used which is the combination of the issuer
field and the serialNumber field from an X.509 certificate.  Likewise,
this same identifier is used for identifying a certificate that should
be used for signature verification.  This identifier is not
suitable/applicable for non-X.509 environments.  Also the certificate
shuttling mechanism is based on an ExtendedCertificateOrCertificate
which is an X.509 certificate, or a (deprecated) PKCS #6
ExtendedCertificate -- not really easy to put in something that isn't

My understanding is that PKCS #7 is being worked on to address some of
these things.  SMIMECapabilites is helpful for supporting different
public-key models, but I think there's some more work that needs to be
done in order to have full support.