Re: [idn] WG last call summary

[Paul Robinson <paul@iconoplex.co.uk>]

On Mar 19, "D. J. Bernstein" <djb@cr.yp.to> wrote:

> Go sell a Greek user an ``internationalized domain name'' with a delta,
> Pete. Then tell him that most of his correspondents will see the delta
> as incomprehensible gobbledygook rather than a delta. See what he says.

OK, scenario 1:

You tell him that although it's gobbledygook to people without greek
alphabet support, it will still work. It's not convenient, but it WILL work.
Guaranteed. For his business colleagues and friends in Greece, who DO have
the latest and greatest software, it will display as a delta. His ISP hasn't
had to upgrade, and everybody in the world can use his domain - eventually
they will see it as a delta as well, but for now they see it as an encoded
string they can still use no problem.

Scenario 2:

Oops, sorry, our mistake, it's NOT gobbledygook, it's prefectly fine. For
everybody in Greece. Unfortunately, his bank in the UK can't understand his
e-mail address because the S/360 coders haven't got time to upgrade all the
systems and applications software. His family won't be able to send him mail
through systems that are running proprietary or legacy mail applications
because they don't understand this 8-bt stuff. When he's abroad, his website
and e-mail address may be useless. But it's OK, because it's a CLEAN
implementation and a great protocol, and everybody else will catch up
sometime in the next 4-10 years. Until then, he has to get a 'normal' domain
to see himself over.

> Of course, display failures are not as intolerable as interoperability
> failures. But they're still failures.

And they are failures for OS developers and application developers. Not the
IETF. Not for the IDNA WG. Not for anybody who wants to get IDNs through.
Not for the people who don't want to have to re-write the MTA on the PDP
they have running in the back office. Not for people who want to have to
deal with another SMTP spec change. The only problem as I see it, is that
until software that deals with IDN knows how to display PunyCode properly,
people will see some crap on the screen. What you are proposing IS
introducing an interoperability failure, which through your own admission is
worse than a display failure.

> Surely you agree that bounced mail is serious!

Which of these is easier to implement:

1. An updated DNS resolver
2. Making every piece of software and display device that might ever have to
deal with IDNs capable of handling UTF-8?

If you were IT director of a large firm, and you had a choice as to which to
roll out, which would you choose?

--
Paul Robinson