MD2 is is known to have pseudo collisions. In the previous version of
the draft md2 was a SHOULD (along with the rest of the RSA algorithms).
You have promoted it when I consider it to be a suspect algorithm.
There is a difference between what we consider to be good practice and
how backwards compability works. I think that making it a MAY (or
omitting entirely) and adding a note that this is a common algorithm
still (is that really true? 11 out of 108 with what types of experation
dates) is sufficent to address your needs without having an endorsement
of this as a good algorithm on our (the WG) part.
Remember that md2 is not an acceptable PKIX algorithm. I think we
should follow that lead.
From: Blake Ramsdell [mailto:blake(_at_)brutesquadlabs(_dot_)com]
Sent: Monday, March 18, 2002 3:10 PM
To: jimsch(_at_)exmsft(_dot_)com; ietf-smime(_at_)imc(_dot_)org
Subject: Re: I-D ACTION:draft-ietf-smime-rfc2632bis-00.txt
Thanks for the comments, Jim -- one quick question below.
----- Original Message -----
From: "Jim Schaad" <jimsch(_at_)nwlink(_dot_)com>
To: <ietf-smime(_at_)imc(_dot_)org>; "'Blake Ramsdell'"
Sent: Monday, March 18, 2002 10:27 AM
Subject: RE: I-D ACTION:draft-ietf-smime-rfc2632bis-00.txt
1. I strongly disagree that md2-with-RSA is a MUST. I think this
should be a MAY or omitted.
On what basis you you disagree?
For compatibility, dropping MD2 may not be the best idea.
Based on a quick
evaluation of the root self-signed certificates that I have,
I found 108
total certificates, 11 of which were signed with MD2 (44 were
MD5, the rest with SHA-1).