Re: nl_langinfo(CODESET) again

[Bruno Haible <haible@xxxxxxx>]

Edmund GRIMLEY EVANS writes:
> In glibc-2.2, will the charset returned by nl_langinfo(CODESET) be a
> MIME-compatible name such as "us-ascii" or "iso-8859-15"?

No. The value of nl_langinfo(CODESET) is system dependent and not
intended to fulfill MIME standards. Some of the values happen to
coincide with the corresponding MIME name, some don't. Here is the
list I made for glibc-2.1. glibc-2.2 will support many more locales,
but you get the idea.

       name              a MIME name?
   ANSI_X3.4-1968            no
   ISO-8859-1                yes
   ISO-8859-2                yes
   ISO-8859-5                yes
   ISO-8859-7                yes
   ISO-8859-8                yes
   ISO-8859-9                yes
   ISO-8859-13               no
   KOI8-R                    yes
   KOI8-U                    yes
   UTF-8                     yes

> I recently put code into Mutt to use nl_langinfo(CODESET) for the
> default charset, and people are already complaining about ill-effects
> caused by "ISO8859-15", for example.

"ISO8859-15"/"ISO-8859-15" is not a MIME registered charsets. You
cannot send out a mail in this charset. I'd recommend to send it out
as US-ASCII if it contains only ASCII characters, and UTF-8 (or UTF-7)
otherwise. Mailers are supposed to support UTF-8 for 1.5 years now.

Bruno
-
Linux-UTF8:   i18n of Linux on all levels
Archive:      http://mail.nl.linux.org/lists/