Dear Draft Authors:
The Working Group Last Call is almost finished (it closes on Monday). I
wanted to post the things that I noticed in the document that I have not
seen posted by anyone else.
In section 2, the document says:
The identifier's parameters field contains the initial
vector IV as an optional parameter.
IDEA-CBCPar ::= SEQUENCE {
IV OCTET STRING OPTIONAL -- exactly 8 octets }
If IV is specified as above, it MUST be used as initial vector. In
this case, the ciphertext MUST NOT include the initial vector. If
IV is not specified, the first 64 bits of the ciphertext MUST be
considered as the initial vector. However, this alternative of not
including the IV SHOULD NOT be applied in S/MIME.
First, please change "initial vector IV" to "initialization vector (IV)" or
"initialisation vector (IV)" depending on you geographic preference (US
English vs. UK English). Then, use IV throughout.
Second, we have already seen messages requesting the removal of the
SEQUENCE wrapper in the ASN.1. This should be done. The OPTIONAL is not
needed either. In the AlgorithmIdentifier structure, the parameter is
already OPTIONAL. I suggest:
IDEA-CBC-IV ::= OCTET STRING -- exactly 8 octets
Third, I would like to see the final sentence in the last paragraph
reworded. I suggest:
However, the IV MUST be included in the AlgorithmIdentifier parameter
when IDEA is used with CMS.
Later in section 2, the document says:
The identifier's parameters field MUST be NULL.
Many algorithms use this technique. Since the AlgorithmIdentifier
parameters are OPTIONAL, the same semantics can be provided with fewer bits
on the wire by requiring that the parameters field be absent. Please
consider this alternative.
In section 3.1, step 3, there is a typo. Please change ":=" to "=".
Thanks for your attention,
Russ