baking into the protocol

[John C Klensin]

--On Wednesday, 20 December, 2006 14:38 -0800 Erik van der Poel
<erikv at google.com> wrote:

> Of course, it would take a while to come up with detailed and
> complicated "script" mixing rules. All I'm saying is that we
> could
> simply prohibit some simple (Unicode) script mixtures for now,
> and
> then come up with more detailed rules later. But I'm guessing
> that
> your position is that such a set of rules would fragment the
> DNS
> namespace permanently(?). Did I get that right?

Unless we were more clever than I think we have figured out how
to be so far, yes.

More generally, I think we are at some risk of simply killing
IDNs as a useful concept (independent of whether they are useful
politically or as mechanisms to feed more money into registrar
or registry pockets).  For example, if we push the issues out to
the applications implementers --the browser vendors in
particular-- and they take the position that punycode should be
displayed on the slightest provocation, I think two things
happen, at least on a global basis.  One is that registrars will
have little to sell, because they cannot guarantee that IDNs,
once sold and registered, will ever be displayed.  The second is
that users will be so used be being "warned" about
supposedly-dangerous IDNs that they will ignore all such
warnings as not containing any useful information about what is
and is not safe.    The combination could leave us with a
situation in which IDNs get a lot of use in ccTLDs in
few-language countries but otherwise are not fully functional
for a large proportion of the Internet's domains or users.

     john