Re: Variant tests failing on big endian machines

[Emmanuele Bassi <ebassi gmail com>]

Hi;

On 28 June 2015 at 22:09, Torsten Schoenfeld <kaffeetisch gmx de> wrote:
> On 26.06.2015 17:54, Rafael dos Santos wrote:
> > > const char * g_variant_classify (GVariant *value);
> > >     PREINIT:
> > >         GVariantClass vclass;
> > >     CODE:
> > >         vclass = g_variant_classify (value);
> > >         RETVAL = (const char *) &((char) vclass);
> >
> > GVariant.xs: In function ‘XS_Glib__Variant_classify’:
> > GVariant.xs:285:26: error: lvalue required as unary ‘&’ operand
> >   RETVAL = (const char *) &((char) vclass);
> >                           ^
>
> Right.  How about this patch then?  Works on my (little-endian) machine.
>
> (The "const char *" typemap copies the string, so it's OK to return
> references to local variables.)

Can't we just return a char? The GVariantClass enumeration type only
contains the ASCII value of the *first* element of the type string.
You're supposed to switch on the value, and recurse if the type is a
complex one.

It's literally the equivalent of:

  return g_variant_get_type_string(variant)[0];

We can either do:

char g_variant_classify (GVariant *value)
    PREINIT:
        GVariantClass vclass;
    CODE:
        RETVAL = (char) g_variant_classify (value);
    OUTPUT:
        RETVAL

Or we could return a copy of the type string, limited to the first character:

char * g_variant_classify (GVariant *value)
    PREINIT:
        char *res;
    CODE:
         RETVAL = g_strndup (g_variant_type_string (variant), 1);
    OUTPUT
         RETVAL

If we still want to return a string.

Ciao,
 Emmanuele.

-- 
https://www.bassi.io
[ ] ebassi [ gmail com]