Re: one-language solutions, localised DNS and internationalised DNS?

[tinwee@pobox.org.sg]

"Martin J. Duerst" wrote:
> At 09:38 00/01/24 +0800, James Seng wrote:
> > I asked about one-language solutions because I know of many attempts to do a
> > localized DNS. Localized DNS in the sense that they make the DNS 8-bit cleans
> > and then used localized encoding in the zones. This is probably the simpliest
> > and most obvious solution but of course it has a *lot* of problems.
> 
> I see. Calling them localized DNS may be okay. Calling them one-language
> DNS is confusing, because language is not really the issue.

We met last week to talk about localised DNS and internationalised Domain
names, and reviewed the impending proliferation of solutions to the question of
multilingual domain names. There is an urgency to work out the IETF draft
quickly 
as many folks are coming up independently with localised solutions with 
one-language support, one-encoding support, or various combinations etc. 
and need the IETF draft as a guide.
> 
> > One of which is interoperability, ie how do we ensure a ISO-2022-JP DNS able
> > to interoperable with EUC-JP DNS, or GB DNS and BIG5 DNS. How do the zone file
> > match, how to get unified results etc.
> 
> These are experiments that just went the short path. The sooner we come
> out with something that clearly defines a standard, the less danger
> for those intermediate solutions to persist. As your work in Singapore
> has shown, it is indeed possible to use an intermediate resolver/proxy
> to deal with some encoding issues, but if we ever are going to describe

The different encodings will not go away quickly. They are likely to
persist as legacy problems.

With Tamil for instance, we had initiated an international conference in 1997
to trigger the Tamil diaspora to achieve the challenge of a single 
unified encoding system for their language. Now the Tamilnadu state government
in India is likely to lead the unification attempt on dozens of competing
encodings developed by software vendors, through the offer of standard
Tamilised domain names if they can all agree on a standard. Even then,
I am not sure we will succeed to reduce them to one single encoding scheme
even if it is unicode. In the case of Japan and Korea, especially
if it is Unicode, there is a strong sentiment against it there.

> this, I think it should be done in terms of how to move to a worldwide
> solution as quickly and smoothly as possible, and not in terms of how
> these things can stay around for as long as possible. We definitely
> do not have a requirement to guarantee that any of the current
> implementations, be it on local dns or on how browsers treat
> i18n names, continue working. If we happen to figure out that
> what they do is most reasonable anyway, or we can show them
> a way to converge, that's a plus, but not a must.

I agree with your comments on the need to move to a worldwide solution,
quickly and smoothly. Otherwise, the mess will be difficult to solve.
For example, we just heard that in China, CNNIC is proposing the use 
of multiple roots to solve the multiple encodings and this may lead to 
one root for each encoding. They have gone ahead to announce the issuing 
of flat top level domain names for their own encodings last week. They
are getting TWNIC to form another root for another encoding.
We were also told by CNNIC staff that they are issuing flat Chinese domain
names at top level for each registrant, in addition to hierarchical ones
although I may be mistaken (I copy this to Prof Qian of CNNIC for his
verification).

The work of this IETF WG is absolutely critical to steer the world from
impending confusion by providing clear guidelines for research
and development and sound implementation.


> Regards,   Martin.
> 
> #-#-#  Martin J. Du"rst, World Wide Web Consortium
> #-#-#  mailto:duerst@w3.org   http://www.w3.org

bestrgds

Tin Wee
--
Tan Tin Wee, National University of Singapore
tinwee@pobox.org.sg