Re: [idn] Time to reconsider

[Keith Moore <moore@cs.utk.edu>]

John,

> My personal response would be that this is more evidence that
> trying to do these things in the DNS is a mistake.  The issue
> isn't that of one coding versus another.  It is that we are
> expecting a lookup-and-exact-match system to do the impossible
> -- fuzzy matching that doesn't astonish a user whose cultural
> and linguistic assumptions cannot be predicted or identified
> given the tools at hand.

Here's why I'm skeptical about this argument.  Most of the 
computer operating systems I'm aware of use more-or-less
the same matching algorithm for filenames (including filenames
typed by humans) as that used by DNS - exact comparison modulo
case folding.  Yes, point-and-click interfaces exist, but they 
don't scale.

Now it might turn out that that kind of filename matching is 
completely impractical for some language because they have
too many different ways of spelling words.  But I suspect 
that, to the extent possible, folks who speak languages
that have multiple ways of spelling words can develop
conventions for spelling domain names in that language - just 
as we have found some restrictions on English-language domain 
names (e.g. the LDH rule) useful in minimizing the differences 
in how domain names can be spelled.

I do think we should realize the limitations that are inherent
in the approach of extending DNS to non-ASCII characters.  
However, I don't think we should stop developing an IDN solution
simply because it's not the perfect solution to all name lookup
problems.

The other bit is that it's very hard to build a federated name
space once you get away from canonicalization followed by exact 
matching.  Which means that in such a world you either have a 
single point of control for ALL names (not just the TLDs), or 
you don't get consistent results from one lookup service to 
another.  Neither one of these sounds particularly desirable as
a replacement for domain name lookup.  The struggle for control
of the DNS root and TLDs has been too painful already, and the
major contributors to that war have shown no sign of willingness
to exercise restraint or minimize the collateral damage.
 
The last thing we need is to have a fight over control of all names.
Because people will demand fuzzy-matching all the way to the root,
with all of the chaos and inefficiency that this implies.

So I think that IDN is still needed, even though we may also 
recognize that a fuzzy-matching lookup service might also be
needed.

Keith