Re: non-ASCII characters in /etc/locales.alias ?
At 24 Jan 2002 11:48:03 +0000,
Alastair McKinstry wrote:
> (Final?) proposal:
> (1) Documenting locales.alias man page, saying don't add non-ASCII
> elements. In fact, preferably don't add any. Explain why.
> (2) 'locale -a' displays only usable aliases; none that can't be
> displayed in the current locale or don't have correct locales generated.
> If there are no comments by Monday, I'll submit the bug reports and
> write the patches,
The document should discourage usage of two non-ASCII locale aliases,
because these locale aliases are left unremoved just because of
compatibility for older systems. Especially, the usage of these
locale aliases are invalid in C (or POSIX) locale and multibyte locales
(such as UTF-8, EUC-JP, and Big5). In fact, new users have no reason
to use these locales.
A line of "# Read locale.alias(5) before edit this file" in locale.alias
file. (In truth, locale.alias(5) says that this file should not be
edited, as Alastair says.)
BTW, do you think usage of these locale aliases in, for example,
KOI8-R locale is valid? If these aliases should be regarded as
pure byte sequence without encoding (even if users feel impression
that the byte encoding is a human word), the answer would be yes.
If these aliases imply certain characters, the answer would be no.
Tomohiro KUBOTA <firstname.lastname@example.org>
"Introduction to I18N" http://www.debian.org/doc/manuals/intro-i18n/