This is exactly what frustrates me the most and what most of my logic checks. I'm writing tests for each of these semantic cases, but it seems silly to me that a standard leaves something so ambiguous.