Hacker News Reader

Nice trick. One thing bothers me, couldn't the compiler use operator float and the construct from float for non-matching units?

I believe using those together counts as two user-defined conversions and is rejected. But one or both of those might as well be declared explicit.