Parsing the issues and finding a middle ground -- another attempt

[John C Klensin]

--On Thursday, February 26, 2009 14:02 -0800 Erik van der Poel
<erikv at google.com> wrote:

> I guess that depends on what would be dropped on the floor. At
> this point, some may wish to drop Eszett, Final Sigma, ZWJ and
> ZWNJ so that we can at least get IDNA200X out the door with
> the most recent Unicode version. Do we need another consensus
> call on these characters?

FWIW, I believe that having to revisit every consensus call
because someone says, e.g., "I realize that this would mean a
number of things the WG would like to solve will hit the editing
room floor" would be an extremely efficient way to be sure that
we would never converge on anything.  That was certainly not the
intent of my note and I hope no one [else] reads it that way.
Note that while one might claim limited communities of interest
for Eszett and Final Sigma (although I would not care to do
that), the population of folks who cannot sensibly write large
numbers of words (or mnemonics based on them) in their languages
without ZWJ and/or ZWNJ is somewhat above a billion people.
I'm not ready to blow them off and hope no one else in the WG is
either.

> Responding to John's email, I think it might be best for
> IDNA200X to avoid any MUSTs/MUST NOTs related to mapping. The
> mapping issues should probably be left to another document.
> This means that that other document decides whether to do any
> mapping for characters beyond Unicode 3.2.

That is, of course, where we started.  But the notion that
mapping was to be a local decision is, I believe, a large
fraction of what has held us up for months.   Part of that is
that "another document", if not either part of the IDNA2008 set
or normatively referenced from them will be adopted by some and
ignored by others as a standard to which they don't have to
conform in order to conform to IDNA.   And that gets us back to
exactly where we were when we started.

I agree about the MUST/MUST NOT language, but only because it is
fairly easy for me to imagine cases that would make anything
that was specified unnecessary or unwise.  But for those who are
concerned that the use of words like "local" justifies rampant
interoperability, SHOULD language in the IDNA specs would seem
to be necessary.

Or do I not understand what you are suggesting?   I know that I
don't understand what Leslie is suggesting because I don't see
anything I wrote as justifying abandoning any fundamental
IDNA2008 goals.

    john