Robert Kawulak skrev:
2008/12/11 Thorsten Ottosen <thorsten.ottosen@dezide.com>:
We need to make either x or y one of the bounds of our interval. Let us assume it is y.
Furthermore, we know y has been specified such that it is representable exactly in float or double.
Not necessarily:
typedef bounded<float>::type bfloat;
bfloat ouch( sin(x), bfloat::constraint_type( cos(y), tan(z) ) ); ouch = log(7);
Any of the bounds or the underlying value may be extended during computations involved in both the construction and the assignment (if I understand correctly), there's no guarantee that the compiler will not optimise away storage of the values.
Well, in most of my use cases the bounds are defined at compile time. As for your example, then I hope we find a way to force this rounding.
And actually what "truncation" is -- really a truncation (rounding towards 0), or rounding to the nearest representable value?
I think it is usually rounding towards nearest. -Thorsten