Archaic scripts (was: Re: New version: draft-ietf-idna-tables-01.txt)

[John C Klensin]

--On Thursday, 08 May, 2008 00:03 +0200 LB <lbleriot at gmail.com>
wrote:

> If I understand your internet engineers debate, your discussion
> suggests that you consider that your proposal may be
> jeopardized by 5000 years old characters. If so, why would it
> not vulnerable to the characters we will need in two years?
> All we know is that classification systems historical (museum,
> university libraries, research, etc. ...) who need a coherent
> system providing universal interoperability between IRI,
> langtags, domain names networks used in translation, and so
> on. will decide to develop their own solution.
> 
> Perhaps it is a good two-tier approach? Thus, there will be a
> limited system proposed by the IETF and a generalized system
> that can gradually emerge in MLTF. The only constraint is to
> keep the same punycode for interoperability of domain names
> with the prefix "xn -".

> I also understand that this is only the BCP for TLD managers,
> as the concerned scripts are always supported by Unicode and
> punycode. This enables the mobile phones to be simplified
> without infringing IDNA?

There is only one domain name system and only one IDN protocol
for the DNS.  Whatever is done for IDNs "in the IETF" will
inevitably be the same as whatever "MLTF" needs to use.  That
said, you have identified a very fundamental issue which some
people keep forgetting:

IDNs form part of the foundation for internationalizational and
localization work in the Internet.  More complex identifiers
such as IRIs and human-friendly identifiers developed in and for
locally and culturally-specific contexts build on that
foundation but are not intrinsically part of it.  IDNs (and
IDNA) are strictly concerned with domain identifiers ("names").
They have nothing to do with what can be written in an email
message, what can be placed on a web page, what can be
transmitted from or read on a mobile phone as text messages, and
so on.   Those types of functions are constrained only by
Unicode if one chooses to use Unicode -- most, although not all,
of them, do not even require the use of Unicode but permit one
to identify other character sets.

Nor is any of the current work about a BCP for TLD managers.
Anything we do here applies equally to all domains in the DNS.
Far  more of those domains are _not_ top-level ones than are at
the top-level.  And, while recommendations to domain
administrators may emerge as part of the work, they are not its
primary objective.

Put differently, some of the issues that MLTF wants to address,
at least as I understand them, build on IDNs but do not need to
be --and should not be-- part of them.  Far more of the MLTF
issues do not involve IDNs at all.   To confuse IDN work with
those MLTF agendas involves what, in networking contexts, is
called a layering violation.  In more ordinary language, it is
just confusion.

    john