[COSE] Review of draft-ietf-cose-hash-sig-01

[Jim Schaad <[e-mail]>]

First, I don't think I have any technical issues with how to use this with
COSE, that said I find there are numerous issues about the description of
HSS/LMS.

1.  Section 2.1: Same comment as for LAMPS about stand-along tree
definition.

2.  Section 2.2: LMS systems have three not two parameters, Hash Function, #
of bytes used, tree depth.  It is possible to use fewer bytes than the hash
function generates.  I don't know that this would ever be used but it is
more correct.

3.  Section 2.2: The registry would also allow for new hash functions to be
used as well.

4. Section 2.3: See comment #2 about adding hash function

5.  Section 4.1:  In paragraph 3, implementation currently violates the
conditions in this paragraph.  I am resigning the same input with the same
key multiple times because I need to generate the signature on the public
key each time I build a signature.  Should this be restated to allow for
that to occur?

6. Section 4.1:  I think that it might be worth while having a statement
about the use of appendix A in [HASHSIG].  The use of this does two things:
1) it decreases the size of the private key and 2) it decreases the
effective size of the private key down to the seed value rather than having
all of the LMOTS keys generated independently.

Jim