Re: Unicode, Cure-all or Kill-all?

From: Jonathan Rosenne (100320.1303@CompuServe.COM)
Date: Sat Aug 10 1996 - 17:24:45 EDT

Michael Everson wrote:
>At 01:14 1996-08-10, Jonathen Rosenne wrote:

>>A standard should be based on existing technology.
>But ISO10646 and Unicode have, since their inception, been expanded, and
>as-yet-unimplemented schemes have been standardized.

By technology I did not mean all the details. The technology in Unicode includes
16 bit characters, decomposition of composed characters, automatic shaping of Arabic
and Han unification, all of which have been implemented and tried. Adding
characters according to the accepted technology is what I call "based on existing

No one has implemented the 75 thousand Chinese characters Prof. Huang mentioned,
no one has any expreience in how to input them or use them in current technology.

And only if the Chinese people themselves do it, get it out to the public, have
it tested by thousands of users in real life and make it a national standard, after
that it would be appropriate for international standardisation.

The same applies for all complex languages - it is not reasonable to expect the
international community of specialists to come up with reasonable solutions for
our problems. The first drafts of 10646 on Hebrew, produced without Israeli
participation and without knowledge of existing technology are a case in point.

And currently, there is no implementation of Unicode with Hebrew - there are only
a few partial implementations.

>>People should not expect standards to solve technology problems. Standards
>>should be aimed at standartisation problems, i.e. when several products
>>implement similar technologies in incompatible ways.
>I disagree. If we can get standardization working efficiently enough, we
>could (in some fields) lead implementations.

This approach does occur from time to time is causes many problems and sometimes
fails the resulting standards. Some of the problems in getting Unicode to be
more widely implemented are related to this phnomenon. It is relatively easy,
for instance, to add a character to 10646, and then let the implementers worry
about obtaining fonts and conversions to existing codes, but they will not do
it unless they have a business justification.

Michael Everson, Everson Gunn Teoranta
15 Port Chaeimhghein =CDochtarach; Baile =C1tha Cliath 2; =C9ire (Ireland)
Guth=E1in:  +353 1 478-2597, +353 1 283-9396
27 P=E1irc an Fh=E9ithlinn; Baile an Bh=F3thair; Co. =C1tha Cliath; =C9ire

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:31 EDT