Re: Unicode and GNUstep (more info)

[Andreas Hoeschler]

Hi Richard,

> 1. On my copy of GNUstep (Debian, Intel), the above test program prints 
> 'char: 246' as
> it does on your MacOS-X system.

Oh!

> 2. What you are doing is actually illegal ... the contents of a 
> constant literal string created using
> the @"" syntax are supposed to be ascii (ie 7-bt characters with the 
> eighth bit clear).  The behavior
> when you use non-ascii characters is undefined.

I always do this on MacOSX and it always works. What would be the 
alternative approach if you want to create a qualifier with fixed values 
in code?

> That being said ...
>
> The ö character has the value 246 in both unicode and ISO latin 1
> The GNUstep default character encoding for C-strings is Latin 1
>
> So it should (and on mys system does) work as you are expecting as long 
> as you have not changed
> your default encoding (GNUSTEP_STRING_ENCODING environment variable), 
> assuming you are using an
> up to date copy of GNUstep of course.

I am using the latest cvs version. I am not aware of having touched 
$GNUSTEP_STRING_ENCODING. However, it does not seem to be set on my 
system.

bash-2.03# echo $GNUSTEP_STRING_ENCODING

bash-2.03# defaults read NSGlobalDomain
NSGlobalDomain Local Time Zone 'Europe/Berlin'
NSGlobalDomain NSLanguages 'English'
bash-2.03#

Is this ok? Where should I set it and how?

> So, if the GNUSTEP_STRING_ENCODING setting is not your problem, and you 
> are using up to date
> software, it would be interesting to know about your machine/os, 
> compiler etc.

I am on Solaris 8 with gcc-3.0.4 compiled from sources.

Thanks,

   Andreas