🚀 go-pugleaf

From a Japanese perspective:

In article <1i99qfEINN420@uni-erlangen.de> unrza3@cd4680fs.rrze.uni-erlangen.de (Markus Kuhn) writes:

   First about keyboard drivers:

	[details omitted]

   Would be a nice research project ...

Agreed. It would not be impossible, but it would require rewriting of
many applications (or input libraries) and would take a long time to
see such changes done everywhere.

   And about sorting rules:

   As I believe that no one will die if his national sorting rules
   are changed, perhaps it's time to create international ISO sorting rules
   comparable to those defined in DIN 5007. Why shouldn't the people that
   know about these problems at first try to harmonize the rules internationaly
   before they implement complex historically grown sorting rules.

May I suggest that before going into ISO arena, each country (or
culture group) codify the existing methods clearly in writing?  And
please note the use of my plural form of method[s]. I don't believe
there is single satisfactory sorting method in each language.  (I
don't know if there is single satisfactory rule for American English,
but it is not clearly in Germany [I read that some publishers dont
follow DIN sorting rule.]  and other places such as Japan.)

   One first approach for an ISO sorting standard would be for all latin languages
   to order letters with diacritics directly behind their versions without
   diacritics. Other suggestions for international sorting harmonization
   rules are welcome. The result might be a huge permutation table for
   Unicode. The goal should NOT be to try to cover all EXISTING sorting
   rules (which may trivially proofed as being impossible), but to create
   something acceptable for sorting multilingual textes. Good idea?

I doubt if this is a good idea because unless we cover existing
sorting rules (and which are presumably used widely), it would be
useless for many users. Like the German publishers mentioned in this
group, who use a sorting rule different from the one in DIN
standard, there will be people using the existing sorting rule(s). So
why not codify them in freely available (semi-) standard documents?

   This ISO sorting standard should be 100% language independend. This surely
   is possible and as I believe would be very very useful.

I don't think this language independence is possible.  I hate to give
a very specific example, but IMHO, location-specific examples and
counter-examples are a norm rather than exception.  So here it goes.
Look at Japanese text (or names). When text (or names) are written
using Kanji (chinese characters), the information concerning how to
read it are more or less lost! [It is true that most of the time, a
grownup Japanese can guesstimate the correct reading. But there are so
many exceptions to rely on the human skill :-(] If we want to collate
such text (or names) according to how we read them, we have to have
auxiliary phonetical data (probably given in the hiragana or katakana
letters.)  I wonder if you call such sorting environment language
independent. [Now, granted, the sorting of hiragana only data or
katakana data is more or less straightforward.] In any case, the local
telephone companies use a particular sorting method to list all the
names in telephone book. I don't think it is language-independent
method at all.

I believe the correct approach to sorting is to codify the existing
rules of each country/culture/whatever, and then publish them in a
easy to read documents which are understandable by reasonably talented
programmers of other countries. Code examples for each sorting rules
will go a long way. This will help I18N from the bottom.

	[bibliographic sorting example deleted]

   ASCII also doesn't allow direct sorting. You have to do case conversion,
   punctuation removement, etc. before you can arithmetically compare the
   strings. With ASCII, these algorithms are only so simple, that most people
   don't recognize them. 

Unfortunately, some languages of the world require much more complex
preprocessing, or the notion of "sorting" may not apply very well in
the naive sense of the word as perceived, say, American users.
Frankly, I have tough time to explain to overseas visitors how the
names are sorted in the Japanese telephone directory. [It would take a
few minutes in the least and I don't know if the explanation is
understodd at all. I bet you can explain the concept of sorting in
American telephone directory in less than a minute to an intelligent
grownup.]

   With ISO 10646, you need the same methods as with ASCII,
   they are only more complex and require much bigger exception tables. There
   is not fundamental difference!

I doubt if it is just a complexity issue. The whole concept of
language independence sounds like a pipe dream to me. (see the
discussion of Japanese sorting above.) There is nothing wrong about
language-specific solutions IMHO as long as such solutions (adopted
from existing practices) are freely available in understandable
documentation.

 Chiaki Ishikawa,  Personal Media Corp.,  MY Bldg, 1-7-7 Hiratsuka,
 Shinagawa, Tokyo 142, JAPAN. FAX:+81-3-5702-0359, Phone:+81-3-5702-0351
 UUNET: ishikawa@personal-media.co.jp