Chris Angelico <ros...@gmail.com>: > On Mon, May 23, 2016 at 1:19 AM, Steven D'Aprano <st...@pearwood.info> wrote: >> Okay, now I'm confused. How is 1/2 returning 0.5 the language not doing what >> you've told it to do? > > That isn't the problem. With binary floats, 1/2 can be perfectly > represented, so you have no trouble anywhere. The problem comes when > you then try 1/5. What do you get? 3602879701896397/18014398509481984. > Python shows that as 0.2. Then you do some more arithmetic, and the > veil is pierced, and you discover that 1/5 doesn't actually return > 0.2, but just something really really close to it - which it tells you > is 0.2. > > I'm not saying that having 1/5 return 0 is better, but I'd like a > broad acceptance that 0.2 is imperfect - that, in fact, *every* option > is imperfect.
Ah, that reminds me of an ancient joke: Ask an engineer what is two times two. He'll take out his slide rule, quickly move the slider and reply: "Approximately four." I remember learning my first programming language, Basic, when I was 16. One of the very first things to notice was the way "the computer" worked with approximations. Marko -- https://mail.python.org/mailman/listinfo/python-list
