Normative diffs

John C Klensin klensin at jck.com
Tue Oct 7 21:45:53 CEST 2008


Hi.

This is something of a side-discussion on the Tranche 1
consensus call, but an issue we will need to deal with no matter
how it comes out.

There have been several suggestions to move Section 9.1 of
Rationale to Protocol.  Some of those suggestions have been
independent of what we do about "normative" material in the
former (i.e., independent of the Tranche 1 questions).

I'm happy to make the move if that is what people want, but am
concerned about this getting tied up with some other issues.  I
believe that:

	(1) It is important that we have exactly one normative
	definition of IDNA2008, with no extraneous material that
	could be confused with the protocol specification.  That
	suggests that, whatever the diffs are, they are not
	normative.
	
	(2) That is especially important because the diffs, as
	we have seen already in WG discussions, do not cover all
	of the edge cases that might occur because of some
	unusual or unanticipated dependency on IDNA2003/
	Nameprep/ Stringprep mappings.   We are not just
	guessing that they are insufficient to be equivalent to
	the definition that is now in Protocol, we _know_ that
	they are insufficient.
	
	(3) For those who want to keep Protocol as compact and
	free of non-normative material as possible, that goal is
	a case against moving Rational Section 9.1 to Protocol
	or for making it an appendix if it is moved (just as the
	actual tables are an appendix to the Tables document,
	segregating them from the normative rules).

Deciding to move 9.1 out of Rationale and into Protocol because
the section is normative would therefore strike me as a large
mistake.  And, if that section is to be moved out of Rationale,
it seems clear to me that it it should got Protocol (in some
form and location) and not to the separate definitions document
that some of us have been talking about as an alternative.

    john





More information about the Idna-update mailing list