AuthnContext classes and submission vote

["Scott Cantor" <cantor.2@osu.edu>]

As you can read here,
http://lists.oasis-open.org/archives/saml-dev/200412/msg00007.html

Ping found some more problems with the proposed context classes. This is
despite no complaints from SQC on these drafts, so there goes another tool,
leaving, well, none.

I have some ideas on fixing them, but it's going to be a lot of manual
scouring of the files, and testing actual XML documents with parsers before
John or I will say with any confidence that they're correct again. The
changes will be less invasive than the fixes we tried to make, since the
basic approach is correct now, we just have to both fix the bugs and figure
out how to evaluate correctness to some degree of satisfaction for
implementers.

We also feel some discussion with people about the original intent of some
of the classes is important to insure nothing is unintentionally changed.

Therefore, I have voted yes on CD status and public review (of which this
is, I suppose, the first), but no on submission to OASIS. I would suggest
others consider changing their votes to no on the submission ballot.

I believe (and the interop seems to support it) that the rest of the spec is
ready, but apart from decoupling this piece from the rest, I don't have a
better suggestion.

-- Scott