Re: [idn] Re: 7 bits forever!

[Keld Jørn Simonsen <keld@dkuug.dk>]

On Tue, Apr 02, 2002 at 02:24:43PM -0500, Edmon Chung wrote:
> James,
> 
> I think you missed the point completely.  The point, as Keld have mentioned
> is that there are people, who believes that it is beneficial to have UTF8
> option, I being one of them and Keld being one of them, and it is a fact
> that cannot be ignored and should not be ignored.  From my interpretation of
> your previous words, you also believe that it is beneficial to go beyond
> ASCII.  That, is my point and please dont try to confuse the message with
> the means!  Human language is not machine language and you have to
> understand the context before you can get the message.  If you cant
> understand this simple depth and intricacy of human interaction and
> communication, I cant help you.

Well, I have to state that I am not so sure we need an UTF-8 upgrade.
Maybe ACE is enough. Things can get complicated if eg the primary DNS
handles utf-8 while one of the secondaries do not. 
DNS is different from SMTP as all DNS clients need to *understand*
all DNS servers, each character needs to be understood. In SMTP the
MTAs are only exchanging messages between consenting adults, and
it is up to the sender and receiver to mutually understand eachother. 

Also the situation back in 1991 with SMTP was different from
todays DNS issues. Then we had a lot of email that was not
ASCII, but a myriad of other encodings, incl national 7-bit variants,
japanese, and iso-8859-?. Today we almost entirely still have
only (restricted) ascii in the DNS.

I frankly think that for interoperation, all DNS data needs to be
stored in ACE encoded ISO 10646, and on the wire it should be ACE
encoded queries and responses. It would then be up to the clients
to do the normalization and conversion from/to ACE.
That would mean no changes to existing DNS protocols!

Then you could have tools for maintenance of the zone data in
one or more charsets for convenience of the DNS zone maintainers.

A weak point of mine is that *iff* we go beyond ASCII, UTF-8 is
the only additional encoding we should consider for the wire,
and *iff* seemless interoperability can be proven with existing 
operations and some gains can be shown, I would not be against 
a new UTF-8 protocol for DNS.

Best regards
Keld