Re: [idn] Alternative Solutions

["Adam M. Costello" <amc@cs.berkeley.edu>]

I think I finally understand Edmon's proposal.

In the current IDNA scheme, the only form of the domain name stored
in DNS is the ACE form.  Applications algorithmically convert Unicode
domain names to and from ACE.

In Edmon's proposal, applications still algorithmically convert ACE to
Unicode, but not Unicode to ACE.  For Unicode to ACE, they do a DNS
lookup, which requires that DNS store duplicate information under both
the ACE names and the Unicode names (or appear to do so).

Presumably this is intended to get around the patent.  I don't know if
it does.  In any case, it seems like such a shame to do a DNS lookup
when a local computation would suffice, and get nothing in return for
that extra cost.  If we're going to pay for a DNS lookup on one end, we
might as well pay for DNS lookups at both ends, because then we can get
something in return: meaningful (non-gibberish) LDH names.

AMC