Re: I18N (Multibyte Enablation) of debconf
On Mon, Aug 04, 2003 at 01:10:44PM +0900, Tomohiro KUBOTA wrote:
> From: firstname.lastname@example.org (Denis Barbier)
> Subject: Re: I18N (Multibyte Enablation) of debconf
> Date: Sun, 3 Aug 2003 23:51:23 +0200
> > Did you write something similar in C?
> > It should be pretty useful, e.g. in cdebconf (a rewrite of debconf in
> > C for the debian-installer, see
> > http://cvs.debian.org/debian-installer/tools/cdebconf/src/strutl.c
> > the strwrap() function) or tasksel (reflowtext() function in strutl.c),
> > and certainly also when displaying translated package descriptions.
> I think it is a good idea. I would like to develop it as a library
> so that it can be used not only from cdebconf/tasksel but also other
> Debian and non-Debian softwares, though I don't know I can do that.
> Anyway, I am thinking about starting a project on Sourceforge.
> (Is Alioth better?)
> In this case, design of interface and encoding system is a problem.
> Which encoding should be used? The encoding specified by the LC_CTYPE
> locale (same as "multibyte character" in C terminology)? Fixed to UTF-8?
> Or, wchar_t? From reading strwidth() in strutl.c which has mbtowc()
> and wcwidth(), I think it is LC_CTYPE encoding.
IMHO you should first focus on providing working solutions for each
individual program, and see later how to provide reusable code.
The main reason is that this would allow Japanese translators to
work on these programs, maybe they are not willing to do so now.
There are surely other problems which are hidden now, and the sooner
these programs are known to work with your locales, the better.
About cdebconf, a fake C.UTF-8 locale is used, so all strings are UTF-8
encoded. Language selection is done via the LANGUAGE environment
With tasksel, LC_CTYPE encoding is used.