The lookalike problem(s)

[Michael Everson]

At 17:18 -0500 2006-11-26, John C Klensin wrote:

>IMO, unless we find a mechanism that no one has been able to
>think of yet, the only "script mixing" rule that belongs in the
>protocol itself is one involving mixed LtoR and RtoL substrings.
>Personally, I wish we could get rid of that one too, just to be
>completely consistent about what belongs in the lookup protocol
>and what is properly a restriction at registration time, but I
>don't see any way to do that without introducing far more
>confusion about ordering and rendering than the principle could
>possibly be worth.

I don't understand any of this. Once I thought we had been making 
progress and agreements. Now all I see is this entire endeavour 
heading for disaster, because whatever progress and agreements were 
made are now being mooted for jettisoning. I cannot fathom why. John, 
you wrote me privately and said that people were considering this an 
engineering exercise. You wrote a long and considered post to me, and 
I owe you a considered response, which I haven't made because I was 
travelling this weekend. But when I saw what you've written above... 
well, I'm just gob-smacked.

Language is not engineering. One cannot algorithm one's way to 
neatness in this effort. Language isn't tidy.

John? If we do not ban script mixing, how do you propose that IDN 
will work? I mean, how?

Maybe I'm wrong. Maybe going back to first principles makes sense and 
will help. But I do not understand how rescinding a ban on script 
mixing can possibly help make IDN a reality.


-- 
Michael Everson * http://www.evertype.com