Good comments for the list,
I suggest a "Capabilities Attribute" that's self-signed and dated to notify
the other party of changes (sound familiar?). I could not care less if
someone wants to use patented $$$ algorithms, as long as it's NOT a "MUST".
OIDs work fine here. PGP 5.x already has this, RSA, DH, DSA, and DES, 3DES,
IDEA, CAST, with MD5, SHA-1.... I think we should have DH with DSA, SHA-1
and either 3DES or CAST as a MUST, all else is fair game, but not required.
PGP Inc does not want to dictate this list of algorithms, that's what the
working group will decide. PS: I'd like to see at least one EC in the list
of "MAY" asymmetric algos.
CB
At 02:03 AM 9/12/97 -0000, iang(_at_)systemics(_dot_)com wrote:
Charles,
Good list.
To which I would add a desperate need: some mechanism in which to
add new algorithms. This is partly technical: the formats of some
generic public key method are obviously going to be tricky, and the
symmetric algorithms require a byte or int allocating type.
It's partly managerial: somebody's got to do the work of holding the
list of numbers, and allocating new ones. A cryptographic Jon Postel
as it where.
And it's partly political: someone has got to avoid the various
rent-seekers and other squeezers of this world, such that a purely
socio-technical benefit is achieved.
Maybe this isn't what you intended for the list, but it is something
that has dogged the members of the Cryptix Development Team in the
past: no way to consistently add a new algorithm. By the end of
the year, we will have more than 10 algorithms that we want to put
into a new version of PGP code. How? A mystery so far.
Whilst on the subject, I'm not sure it is appropriate for PGP, Inc to
perform this function. We need someone outside the US. Also PGP,
Inc and the predecessor team have a record for secrecy that the NSA
would kill for. Not that this is entirely their fault, but history
cannot be changed, and it's simply not helpful to the coder.
--
iang
iang(_at_)systemics(_dot_)com