CJK: Subset of Unicode to represent Japanese Kanji? (was: Thought s)

From: Marco.Cimarosti@icl.com
Date: Mon Jul 17 2000 - 14:39:39 EDT


Michael W. Martin
> For a device that will print a relatively basic label (such
> as sequence
> number, date, time, name, department, etc) onto a document in
> Japanese --
> what is your consensus? Basic Kanji+Hiragana+Katakana.... or will
> Hiragana+Katakana or just Katakana suffice?

My vote is that kanji are definitely needed in a decent application, in year
2000.

If your customers would accept a katakana-only system, I bet that they would
also accept an English-only system... The Hiragana+katakana solution sounds
odd to me (I naively associate hiragana with kanji).

The number field presumably uses western digits. Date and time field only
need a handful of kanji for "year", "month", "day", "hour", "minute", etc.
But these *have* to be used: even if everything else is in katakana, a
Japanese would expect to see these very basic kanji. (E.g., I have an old
Japanese pocket computer whose single-byte character set includes katakana,
a few graphic symbols *and* the dozen-less kanjis needed in dates).

About the name and department fields, you should be more precise:

1) Is the name a person's name (e.g. the customer) or does it come from a
small closed set (e.g. the device's model name, the film type, the
description of requested processing, etc.)?

2) What is the department field: a number or an "alpha"-numeric string?

3) Where do these strings come from (e.g. static text, or typed on the
device itself, or uploaded from a remote computer)?

If people's or places' names are involved, you have no choice: you have to
implement at *least* the whole JIS X 0208 set (and, BTW, this is not even
enough to protect yourself from occasional complaints from people whose name
contain unusual kanji).

Question 3 is particularly important, because if the text is arbitrary and
it is typed on a standard computer, your device necessarily needs to support
the same character set as the host -- unless you also write the host
application and its input method, and can thus impose your limits at the
source.

_ Marco



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:05 EDT