ietf-smime
[Top] [All Lists]

Re: CMS: signed attributes?

1998-03-26 06:15:35
Dave:

With the addition of MACs in the form of authenticated-data, I think that
signedAttributes and unsignedAttributes will make it clear that signed-data
does not rely on anything in authenticated-data.

I will make the change in the next draft.

Russ

At 01:03 PM 3/23/98 -0500, David P. Kemp wrote:

I suggest that the CMS SignerInfo fields "authenticatedAttributes" and
"unauthenticatedAttributes" be renamed to "signedAttributes" and
"unsignedAttributes" respectively.

There are two justifications for this request, one trivial and one
substantive:

* Trivial reason: "signed" is easier to type than "authenticated",
 and it sounds pretentious to use five syllables when one will do.

* Real reason: signing is a mechanical process - attributes are either
 covered by a digital signature or not.  Authentication is a process
 involving policies and procedures - an attribute is not "authenticated"
 unless the signature (including the cert path) verifies correctly and
 the certificate policies allow the use of the attribute.  It is incorrect
 to call an attribute authenticated solely because it is signed, just
 as it is incorrect to call a certificate "valid" because it is signed.


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