On Sun, Jan 11, 2015 at 05:28:30PM -0500, Salz, Rich wrote:
I'd go even further than that and just mandate MUST ASCII.
I don't know if the IETF is "allowed" to do that any more, but +1 for
the reasons you list.
I'm not sure if the IESG will agree to that, and I'm not sure if the
IETF will agree to that (well, this is an IETF LC, so we're finding
out). The shepherd can probably give us a sense of the IESG, and we can
always come back and make it fit what the IESG wants.
As for myself, I'd rather we do something like this:
- note that PKCS#11 used to be just-use-8
- note that what used to interop did because of local conventions (what
locales systems and/or user sessions use)
- note that PKCS#11 _now_ says use-UTF-8 with no further advice
- [therefore] note that
a) US-ASCII is most likely to interop,
b) where non-US-ASCII is needed then UTF-8 is most likely to interop
(and is required by the latest PKCS#11 spec),
c) where UTF-8 is used then one should normalize to NFC on create and
lookup (and possibly use form-insensitive string matching).
I can't see the harm in any of those. Requiring US-ASCII when PKCS#11
doesn't does strike me as a bit too pessimistic, but I'd be fine with
RECOMMENDing it.
Nico
--