Re: [Gnash-commit] gnash ChangeLog testsuite/actionscript.all/ops.as

[Sandro Santilli]

On Mon, Aug 27, 2007 at 11:50:03AM +0800, zou lunkai wrote:
> >  x = new String("a");
> >  y = --x;
> > +check_equals(typeof(y), 'number');
> > +xcheck(y!=NaN); // uh ? is it a different NaN ?
> > +check(! (NaN!=NaN));
> > check(isNaN(y));
> > +check(isNaN(NaN));
> 
> Tests confirmed.
> 
> yes, y might be a differnt NaN(y is an object).
> 
> see added tests below:
> 
> +x = new Number(NaN);
> +y = NaN;
> +check(isNaN(x));
> +check(isNaN(y));
> +xcheck(x != y);      ----->[1]
> 
> As we see,  [1] should return true.  All I can think about is that x
> is an Number object, and y is a primitive number.  Doesn't mean much,
> but might help the valueOf model...

ECMA states that if one of the operand in an equality operation is a number,
and the other is an object, then the object is converted to a primitive
type (in this case, a number) and that value is used in comparison.

Try:

        five = new Number(5);
        check(five == 5);
        check(five != 6);

NaN seems a special case...

--strk;