Re: [idn] DNS is just one of many protocols that use domain names

[C C Magnus Gustavsson <mag@lysator.liu.se>]

On Wed, 20 Mar 2002, Adam M. Costello wrote:

> You are both focusing too much on DNS.  DNS is just one of many
> protocols that carries domain names.  Others include SMTP, RFC 822,
> IMAP, HTTP, URIs, SSL certificates, and on and on...

Yes, I know. Considering how many times this mantra has been repeated
on the list, it seems unlikely anyone wouldn't be aware of it. ;-)

I get the feeling we've been here before. I hope you don't mind me
saying so, but as soon as anyone suggests anything but ACE you claim
this person is focusing too much on DNS. I think we all see the
advantages of ACE, but I'm not so sure everyone see the drawbacks.

> Each one would need to be updated.

Yes, or at least looked over; that's the unfortunate reality. The real
question is whether it's worth it or not, and I suppose that's what
the discussions in this wg with few exceptions have always been about.

I wrote some time ago, when this list was recently set up (and quoting
Linus Torvalds) that when some things need to be fixed, they really
need to be fixed. We have come a long way since but I still see this
as one such case. Perhaps even more so. We then had the choice between
a quick and dirty fix that wouldn't solve all the problems and a more
painful solution that would be better in the long run. We still do.

In this matter, I have to agree with Dan Bernstein, Dan Oscarsson and
others; I still really think we should do it right from the beginning.
We have a five story building on quick sand here and I'm not so sure
we should add another five stories on top of it without partially
tearing down the old one to build a more solid foundation first.

You can hardly claim my suggestion will break anything. It will take
more effort and more time for sure and my suggestion is obviously the
most ambitious. But it will solve the largest number of problems as
well. Again, the question is: Is it worth it? I take it yours and Mats
Dufbergs answer is "no" but I'm not so sure about the rest of the wg.
This is not going to be an easy change-over for the Internet no matter
what solution we choose, so why not do the foundation right? This is,
as said, a golden opportunity to fix a number of other old problems.

> And the harder part of the problem is not how to preserve the
> characters, but how to compare the labels: knowing that e-with-acute
> is the same as e followed by combining-acute.  Simply making protocols
> 8-bit clean doesn't give you that.

Yes. That's exactly my point. This effort has taken many years. By now
we have to realize that it's not going to be an easy solution. So, are
we sure we don't want to do it right? $100 is expensive for a pizza
but when buying a house for $150,000, adding $1000 isn't much.

Magnus