Re: [idn] Cherokee letters look like uppercase Latinletters

[John C Klensin <klensin@research.att.com>]

--On Thursday, 26 July, 2001 17:44 +0900 Soobok Lee
<lsb@postel.co.kr> wrote:

> The initial ascii domain idea  was not a mistake, but a
> practical choice  in so early stage of the internet when  
> we had no standard for UNICODE-like universal character set .

Without getting involved in the other issues in this thread, the
above statement might be historically misleading although it is
correct.  

The original decisions came in  more than one stage, and predated
the Internet by some years.  In at least one of those cases,
internationalization issues were explicitly considered
--obviously in the context of things that evolved into ISO 646
national variants, rather than Unicode-- and the conclusion was
to focus the identifier character set on a set of characters that
was as common, broadly available, and usable at the time.  The
usability discussions considered issues very similar to those we
have discussed, including transcription from other media
(including hand-written text).  

The choice of that ASCII subset was reasonable --for identifier,
not language-- purposes at the time, but obviously is not
appropriate if one is dealing with people or devices that can't
render ASCII characters.

But at no time where non-"hostname" or non-ASCII characters
ignored in the process.  The decisions were made on the basis
that the names were identifiers, that they would rarely be seen
by ordinary users, that absolute precision of the identifiers was
a requirement, and that all users who did need to look at the
names would be able to create and render ASCII-subset characters.
We are having the discussions we are today because changes in
Internet protocols and usage patterns have called some of those
assumptions into question.

    john