We already have the special rule that if either operand of multiplication or division is decimal with variable scale, the result is decimal with variable scale.

We need to add floating scale to the mix.