global locale setting (Re: non-ASCII characters in /etc/locales.alias ?)

[Tomohiro KUBOTA]

Hi,

Since I have not read the main stream of this discussion,
only one point.

At 23 Jan 2002 09:53:04 +0000,
Alastair McKinstry wrote:

> BTW, I like the set-language-env package that you created, and think it
> should be expanded with a debconf-type interface, and if run as root,
> generate the default system locale. Any thoughts?

It would be nice, if a machine is used by only one user or users
who all speak same language.  However, so far, I think it is
technically difficult.  It is because we cannot have ideal settings.
For example, we cannot write Japanese configuration which works
everywhere.  It is about 60% solution.  Thus, I would like users
to read the settings generated by set-language-env carefully and
modify them according to users' way to use.

For example, there are rather small number of terminal emulators
and consoles which can display Japanese.  Especially, the Linux
console cannot display Japanese.  Even though there are softwares
to enable it ("kon" and "jfbterm"), it may be not available according
to the video mode of Linux console.  Thus, LANG variable should not
be set to ja_JP.eucJP in Linux console.  I managed this situation by
checking TERM variable in ~/.bashrc but this is 60% solution.  There
are many terminals and it is not always possible to check Japanese
availability from TERM variable.  This is why I am working on 
internationalizing various terminal emulators.

Another example.  "Window Maker" needs invocation of "wsetfont"
utility to use your language, if your language is not ISO-8859-1
or other simple languages.  It is user-level utility, not intended
to be used by root user.

So far, language setting is very complex and skill-required.  You
can find books how to configure your Japanese environment.  It means
that the amount of configuration items can as big as one whole book.
I think it is a very bad situation, and I am working decreasing
the configuration items, until all I have to do is setting LANG
variable (and maybe input method preference).  (And, at last, these
settings could be done by GUI).  However, this is a far aim and we
of course want to use Linux though the aim is not realized so far.


However, I imagine a "perfect" solution cannot be realized until
far future.  Even Japanese people will say "this is perfect,"
Thai people won't.  When Japanese and Thai people will be satisfied,
then Hebrew people won't.  When Japanese, Thai, and Hebrew will,
Arab won't.  When all of them will, then Mongol won't....

Anyway, I think we need a realistic timeline.  I think that when
Linux console supports most of Unicode BMP characters (except for
typesetting-purpose 1/4 space and so on) for both displaying and
inputting, It will be a time we should proceed.  (Inputting is far
more difficult than displaying, unless we are satisfied by Unicode-
codepoint-input.)  I estimate this situation would be true within
two years.  (One year for displaying and another one year for
inputting).

---
Tomohiro KUBOTA <address@hidden>
http://www.debian.or.jp/~kubota/
"Introduction to I18N"  http://www.debian.org/doc/manuals/intro-i18n/