RE: [idn] pool deadline imminent

["Paul V. Mockapetris" <pvm@a21.com>]

At 10:21 PM 7/16/2001 +0200, Patrik Fältström wrote:
> --On 01-07-16 11.49 -0700 "Paul V.
> Mockapetris" <pvm@a21.com> wrote:
>
> > I believe UTF8 solutions deserve consideration as well.
>
> Please explain what you mean by UTF8 solutions?
>
>   paf

Patrik, Sorry a sentence got lost.

I think of this issue from a broader base than just IDN.  My opinion
is that we need three efforts:

1. A standard for the use of pure binary/8 bit clean labels (and of
course content) in the DNS.  This will require changes.  (for
example, the ASCII matching rules)  A requirement for pure binary
APIs should be articulated, even if the APIs themselves are not. 
BCPs so that ISPs et al don't hurt themselves with shell scripts may be
desirable.

2. Standards preparing, matching, canonicalizing UTF8 at the edge of the
DNS before it is carried in 8 bit clean DNS.

3 Standards for whatever encodings (to build APIs) are wanted by those
that don't want 8 bit clean interfaces.

The intent is that these are layered, so that those that want to bypass
encodings and go 8 bit clean can do so.  The UTF8 standards should
not be intended for all DNS labels; for example if some application wants
to use DNA sequences as four base pairs per byte, that should be 
OK.

I'm really only an advocate of the first two, the third step is to make
this politically palatable 
to those that think UTF8, 8 bit clean, and some rules at the edges aren't
enough.

UDNS seems the closest, although there are things I would change.



paul