in the case of the overflow in the decimal->integer conversion
we didn't return the proper boundary value, but just the result
of the conversion we calculated on the moment of the error