On Wed, 28 Mar 2001, David Goldstein wrote:

> Patent flap slows multilingual domain name plan
> Intellectual property claims have blindsided the Internet Engineering
> Task Force and could derail the group's efforts to develop a common
> scheme for supporting foreign-language domain names across the Internet.
>  http://www.nwfusion.com/news/2001/0326patent.html
> 

I have only had a cursory read of the patent last night but
it's a shame in many ways that the IETF doesn't just
challenge the patent (something they say that they
won't do)

It seems at first blush to be something that the patents
office should have knocked back as being an obvious
application to someone reasonably "skilled in the art". For
some reason they don't seem particularly hard on the
"obviousness" part of the patent awarding process these
days - particulary for software.

Now it is possible I am missing something here as I must
disclose that I haven't given the technical challenge much
thought but my understanding is (and this is backed up the
patent abstract). 

Claim 1. Domain name converted to a standard format that can
   represent all language character sets. eg UNICODE

Comment: Nothing new here just using other people work
and standards.

Claim 2. Then the UNICODE (or whatever format chosen) string
is transformed to be RFC1035 compliant. 

Comment: RFC1035 is the 'Domain Names - Implementation and
Specification' published 1987 (yes 1987!). So nothing
new here either - people writing DNS servers and the like
have been using this document for many years now.

3. How it works.

My understanding (I say that because even though I haven't
looked at the mechanics of this the idea seems fairly
obvious). 

Say some foreign language 'domain' name is represented as
unicode then if the unicode is  converted to some RFC1035
compliant form then the name server can resolve that
just like any other name its presented.


eg Kanji of some name is turned into the unicode string
(looks something like U+11c=222.U+221ee2) - notice we still
have the dot and the 'com' can also be unicode. Now by
simply making this string rfc1035 complaint the name server
will resolve the 'gobbly goop' string it gets presented just
as it does for any current domain name. There's no rule that
says a domain name must be 'someword or name.com' it could
just as easily be 'd-skb0-sw92393dbd.com'

Another way of thinking about this is that you are already
able to register a domain name like 'U-5b66672c'.com and if
'U-5b66672c' happened to be the unicode representation of
'coca cola' in janji then you have already registered the
kanji language domain name. This fact has recenlty been
exploited by people realising how the translation will work
and taking out the appropriate domain names even before it
was officially open for this purpose. 

Anyway back to the patent - they seemed to have patented an
idea and I'd be very surprised if there wasn't an awful
lot of prior art before July 1999 when this was filed.

regards doug


-----------------------------------------------------------
Doug Robb
Clarity Software Pty Ltd      http://clarity.com.au
GPO Box 763                   Phone: 0403 02 2527
Nedlands 6909                 Fax:   (618) 93867564
Australia                     email: doug§clarity.com.au
-----------------------------------------------------------