Re: UCS-2 in gunicode.h

[Havoc Pennington <hp redhat com>]

Derek Simkowiak <dereks@kd-dev.com> writes: 
> 	Sorry, my example was not clear.  The g_print was incidental to
> the fact that the only thing *bytes_converted is good for is error
> checking.
> 
> 	My question remains: g_convert() already does g_warnings followed by
> a returned NULL when the conversion fails:
> 
> [...]
>         g_warning ("Conversion from character set `%s' to `%s' is not supported",
> from_codeset, to_codeset);
> [...]
>         g_warning("iconv() failed: %s", strerror (errno)); 
> [...]
> 

I'm distinguishing these two error cases as "should not ever happen",
vs. the case where the text to be converted is invalid, which can
happen on non-broken systems.

Also, you might want to for example use the text that was converted
successfully, and discard the rest.
 
> 	(Playing the devil's advocate here, I thought "gchar" was supposed
> to be an abstract of the internal representation of a character?  What if
> glib is ported to some Java-like O.S. where a "char" is 16 bits? And
> shouldn't we be using guint8* for UTF-8 input in the g_utf8_*()
> functions?)
> 

No, gchar is simply the same as char, always 8 bits.

Havoc