Hi.
Borrowing from several other notes and comments, it seems to me
that we have three interlocking issues that keep recurring and
producing long discussions. They are by no means independent of
this particular draft, but seem to be becoming generic.
(1) Are we willing to publish (or even standardize) specs whose
nature is to provide a vehicle for making privacy-sensitive
information public? The arguments against doing so seem
obvious. The arguments for doing so include "those who claim
they need this will do it anyway so we are better off publishing
a spec that will at least reduce interoperability side-effects
and permit spelling out the issues as "privacy considerations"
or security ones.
(2) Is turning hardware identifiers (physical-layer objects)
into applications or user level identifiers an acceptable idea?
Are DNS RRTYPEs that map application-level identifiers into
other identifiers that can loop back through the DNS without
guarantees that the process will terminate part of the same
problem or a different one?
(3) Do either of the above answers change if the proposal comes
from another SDO or a major industry group?
I don't know the answers, but I'm pretty sure that trying to
address each of these issues separately every time a new
protocol, RRTYPE, or URN (or URI) type comes along that
interacts with one of them is not the way. It seems to me that
we ought to have something along the lines of RFC 1984 in our
future and that a plenary discussion might be a useful first
step. I don't suppose the IAB or IESG would be willing to
postpone or push something out of the announced agendas to allow
for that discussion, which, given this Last Call and the recent
one over RRTYPs would seem to be critical path. Any volunteers
to get in front of the mic lines?
john