prohibiting previously mapped and unmapped characters

[Harald Alvestrand]

Greg Aaron wrote:
> Dear Harald:
>
> You said: "No sane person would spend the effort to put content on the
> Internet that could *only* be reached by IDNs."
>
> Actually, I know there are very sane people in India planning to use IDNs
> without reliance on LDH domains -- including the Minister of Communications.
> Such use is sometimes a matter of regional or national pride.  And some
> people don't care about LDH names when they have what they think is a better
> alternative.  IDNs are all about providing alternatives.
>   
Note the part about "planning" to use.
At some point in the future (probably after the release of the 3rd 
service pack for IE7), I think providing content solely through IDNs 
will be a viable proposition; my sanity comment (which *is* somewhat 
overblown; I have a tendency to exaggerate) was pointed at the here-and-now.
> People may use their domains differently than you do, and differently than
> you think they will, and for reasons that may be foreign to you.  I simply
> note that due care should be taken when making assumptions about behavior.
> People's choices may not always be wise, or adept.  But it's their choice,
> and choice within boundaries is one of the great things about the Internet.
>
> You do not need to "regard the registry/registrar/registrant relationship as
> the most important part of the Internet" in order to see my point.  The
> point is that if you make a class of existing domain names
> backwards-incompatible, you're affecting their registrants and users, and
> how many of them there are is very relevant and should not be guessed at.
>   
I think it's highly relevant, but I also think we have no choice but to 
guess at them.
> I'll submit again: if one is trying to determine how many potentially
> backwards-incompatible names may be out there, certain approaches are more
> accurate and more objective.  Considering only IDNs that resolve to Web
> sites leaves out a significant percentage of the IDNs in existence, and
> relies on some social assumptions.  I'm glad that you think an examination
> of the published tables at IANA is a good idea.
We have to guess at how big the percentage it leaves out is, too. I'll 
take any metric I can find - and treat it with all due caution, as one 
should always do with any statistic.

                  Harald