Re: Matching and comparison

["Martin J. Duerst" <duerst@w3.org>]

At 14:44 00/01/23 +0800, James Seng wrote:
> Paul Hoffman / IMC wrote:
> > 
> > At 10:01 PM 1/22/00 +0800, James Seng wrote:
> > >I would hate to refer back to any character set even if it is Unicode.
> > 
> > Could you explain why?

> Speaking for my own, and solely for my own, I have an interest to see what
> difficulties and whether a ISO2022-X + ISO8859-X integrated solution (like how
> MULE has done) is viable. I would like to work on this if I have the time...or
> at least see someone else attempt to do it.

Well, MULE has done it, but that fact that it went that way is more
a historical artefact than a very concious decision. They started
their work shortly before Unicode became widely known, and then
always were in a situation that adding yet another feature or hack
was easier at the moment than changing everything to Unicode.
That was strengthened by some very fundamental assumptions in their
basic architecture, which were at least in part due to requirements
from outside (e.g. the requirement for fixed-width cells, and
multiples of the basic width only if you are in a different codepage).
That lead to things such as glyph encodings for Arabic where the
same character can appear on different codepages depending on
its glyph width, which depends on its glyph shape, which depends
on the glyphs around it. Arghhhhhh!


Regards,   Martin.


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