I agree on 1 3 and 4 to be fixed.
I'm not so sure with 2. I think any such recommendations should be
handled in RFC 3851 where the S/MIME Capabilities attribute is defined.
No use of this extension in certificate should differ in any way from
its use as signed attribute in S/MIME messages.
I think that it would be confusing if we started to actually
specify/recommend the attribute content in 2 separate RFCs.
Microsoft Security Center of Excellence (SCOE)
From: Jim Schaad [mailto:ietf(_at_)augustcellars(_dot_)com]
Sent: den 13 december 2004 15:01
To: Stefan Santesson
Subject: RE: I-D ACTION:draft-ietf-smime-certcapa-02.txt
A couple of small comments on the draft, although I believe that it
to last call in its current state.
1. In section 2 you have the statement 'Algorithms should be ordered
preference.' As I general rule I attempt to avoid the use of must,
and may when writing documents to avoid confusion with MUST, SHOULD
(did he just forget to capatilize it?). A better statement might be
'Algorithms are expected to be be ordered by preference'.
2. I would like to see the addition of a paragraph describing the
capabilities that are expected to be listed. It seems obious that
encryption algorithms are listed as, potentially, are key encryption
algorithms (consider RSA-OAEP as an example). However it is not clear
some of the other potential capabililties. What about signature and
algorithms? What about MAC algorithms? What about S/MIME specifics
3. RFC 2199 is a reference, but the text refering to it is absent.
4. RFC 3280 is referenced only from the abstract. Duplicate text
placed in section 1.