The distinction is only relevant for floating point types, is it not?

Yeah, min() returns the most negative int, so it seems like it should do the same for float. Then the float with the least magnitude could have been, say, smallest().