Re: parted nls problem

[Kjetil Torgrim Homme]

Andrew Clausen <address@hidden> writes:

> On Sat, Jul 28, 2001 at 01:04:38PM +0200, Kjetil Torgrim Homme wrote:
> > I think it is a Parted issue, this isn't a problem with any other
> > localised packages.  If I understand the code correctly, it assumes
> > that the output from gettext is multibyte, and that is not the case
> > here.
> 
> It uses mbsrtowcs().  That's all.

Hmm.  Yes.

> > Why are you messing with the strings, anyway?  Rip out that
> > wide character handling, let libc handle it.
> 
> Word wrapping, etc.

Ah...

> The problem occurs when the gettext strings are using a different
> ENCODING (or maybe charset...) to the locale.  IIRC, the gettext
> people said this is a gettext / libc problem, that was going to be
> fixed... but it's a while now...
> 
> Shouldn't mbsrtowcs() be harmless?

I don't know, I don't understand this bug.  I have written an nn.po
using ISO 8859-1.  de.po is also in ISO 8859-1.

 $ LC_CTYPE=de_DE LC_MESSAGES=de_DE sbin/parted 
 Fehler: Kein Gerät gefunden.
 Wiederholen/Retry/Abbrechen/Cancel? cancel
 $ LC_CTYPE=de_DE LC_MESSAGES=nn sbin/parted 
 Error during translation: Invalid or incomplete multibyte or wide character

Why are they treated differently?  And why isn't LC_CTYPE=de working?
It should behave the same.


Kjetil T.