Subject: Re: utf-8 and userland
To: Bill Studenmund <>
From: Noriyuki Soda <>
List: tech-userlevel
Date: 03/13/2004 06:39:32
>>>>> On Fri, 12 Mar 2004 12:07:19 -0800,
	Bill Studenmund <> said:

> Wouldn't nl_langinfo(CODESET) be the right thing to look at?

Applications (e.g. ls) shouldn't use nl_langinfo(CODESET) directly in
this case. It should use iswprint(3).

Also, there isn't such environment variable "LC_LANG".
There are variables like "LANG", "LC_ALL" and "LC_CTYPE", though.

> So I think it's cool, but I don't think UTF-8 an LC_LANG value; it's a=20
> qualifier of how the locale stores characters.

Yeah, "en_US.UTF-8" (or other ${language}_${territory}.UTF-8) should
be used.