Re: [idn] WG last call summary

[Mats Dufberg <dufberg@nic-se.se>]

On Mar 19, 2002, 13:25 (+0900) Bruce Thomson <bthomson@fm-net.ne.jp> wrote:

> > IETF has nothing to do with bugs in programs, just bugs in protocols. You
> > know that. And how long will you plague us with your complaints about
> > sendmail? 10 years? 20 years? -- Stick to what is relevant!
>
> No, this WG has come up with a proposal which is targeted specifically
> at circumventing bugs in legacy implementations.

Dan's complain was about the restriction that mail headers only may
contain ASCII (7 bit). Move that to an appropriate wg.

> If it was only a question of choosing an ideal specification and
> ignoring bugs, "just send UTF" would be a clear winner.

I do not agree. There is a specification of a host name, which sets the
limit on DNS (for host names) to a-z, 0-9 and '-'. The ACE solution is
also better because it breaks less systems during transition.

> Or are you saying that IDNA is needed merely because this WG
> doesn't have access to the particular protocols that need to be
> changed to do this thing right?

We have a reality, and changing to a new model will require a transition.
A good protocol will respect those two factors.


IETF's task is not to repair software bugs or remove limitations set by
the software authors, but when writing new protocol, the IETF has to
respect the reality and to a cerain extent, the effects on currently
installed programs.


If we would create DNS from nothing today, we would do it differently. In
that case, UTF-8 would be a good candidate for the coding scheme. But we
already have DNS, and that fact connot be neglected.



Mats

----------------------------------------------------------------------
Mats Dufberg <dufberg@nic-se.se>
----------------------------------------------------------------------