Re: GNU gettext-0.10.36 for non-ASCII msgid

[Keld Jørn Simonsen <keld dkuug dk>]

On Sun, Apr 08, 2001 at 01:28:12AM +0900, Changwoo Ryu wrote:
> Bruno Haible <haible@clisp.cons.org> writes:
> 
> > It is at ftp.gnu.org and mirrors in gnu/gettext/gettext-0.10.36.tar.gz
> > 
> > 
> > Changes since version 0.10.35
> [...]
> > 
> > * The tools now correctly process PO files in CJK encodings.
> 
> I found that msgfmt complains when it finds a msgid whose character
> sequence is illegal for its po file charset .  For example, nautilus
> 1.0 has msgid _("\t\xB7 Problem : %s\n"), which is illegal for EUC-KR
> charset.
> 
> I think it's gettext bug because the po file charset is for msgstr,
> not for msgid.  But anyway non-ASCII msgid is not a good idea.  It
> sometimes confuses editor, and it is invisible in some locale.
> 
> How about fixing all the msgid's which have any non-ASCII character?

That might be limiting English too much. You have words like naïve
and some words with æ in proper English spelling. And then there
are a number of developers that spell their names with non-ascii
characters. I would myself feel quite offended if my name was
misspelled because of silly ascii restrictions.

On the other hand I see the practicallity of it. The problem
will eventually go away when we shift to utf-8

Keld