-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
I think there's a fundamental mismatch between OpenPGP-style key
preferences and Suite B thinking. As a sender, with labeled
information, you can only use approved algorithms. Thus, if a
recipient
doesn't list the approved algorithm, you just can't send them mail.
The
OpenPGP-style key preferences are in my view primarily to ensure
interoperability and allow for algorithm transitions over long
timescales.
It's not a mismatch, you just have a language that is more expressive
than SuiteB allows.
Are we proposing sender-side rules to match labels to approved
algorithms? It seems inadequate to put 'SuiteBOnly' as a key
preference
on recipients.
No, we're trying to come up with an acceptable protocol.
There are some implementers who are going to want to do SuiteB within
OpenPGP. We should make their lives easy.
But as Werner has noted, it's not our job as an international
community to do things that forbid flexibility. It should be
*possible* to use ECC and Camellia, as I noted in my other message.
The *implementation* may disallow it, but the standard should not. It
should be possible to use ECC and Whirlpool (which people keep
threatening to do a draft for).
A SuiteBOnly pref makes an implementer's job easier if they care about
SuiteB. They can detect it and act accordingly. It doesn't remove the
requirement that an implementer know what they're doing.
Jon
-----BEGIN PGP SIGNATURE-----
Version: PGP Universal 2.6.3
Charset: US-ASCII
wj8DBQFIMhSMsTedWZOD3gYRArkGAJ9IAJUf81i2g6X8dEo8OS+48k2ucwCdHUjy
ULDvJvsrKBfVAXLZMfZhuRc=
=DR2A
-----END PGP SIGNATURE-----