floating point math is math like any other, just with a vastly different number system. There are valid arguments on both side if we should hide or expose those

floating point math is math like any other, just with a vastly different number system. There are valid arguments on both side if we should hide or expose those differences in the implementation, but there's always going to be pretty big implications (i.e. the compiler behaviour mentioned here)