[Python-Dev] Re: PEP 218 (sets); moving set.py to Lib

[Eric S. Raymond]

Guido van Rossum <guido@python.org>:
> > Are you slapping your forehead yet? :-)
> 
> Yes!  I didn't actually know that algorithm.

I thought it up myself.  Which is funny, since to get there you have to
think like a hardware engineer rather than a logician.  My brain was
definitely out of character that day.
 
> This is a generator that yields a series of lists whose values are the
> items of base.  And again, like cartesian product, it's now more a
> generator thing than a set thing.

I don't care where it lives, really.  I just like the concision of
being able to say foo.powerset().  Not that I've used this yet, but I
know one algorithm for which it would be helpful.  Another one I
invented, actually, back when I really was a mathematician -- a closed
form for sums of certain categories of probability distributions.  I
called it the Dungeon Dice Theorem.  Never published it.

> BTW, the correctness of all my versions trivially derives from the
> correctness of your version -- each is a very simple transformation of
> the previous one.  My mentor Lambert Meertens calls this process
> Algorithmics (and has developed a mathematical notation and theory for
> program transformations).

Web pointer?
-- 
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>