[Python-Dev] Re: PEP 276 (simple iterator for ints)

[Nick Coghlan]

(Accidentally sent this just to David instead of py-dev. Although Andrew 
has since clarified that the Set constructor actually accepts an 
iterable directly. And in that case, his magic would probably work 
_regardless_ of how PEP-276 was implemented behind the scenes. Something 
about this aspect of the PEP doesn't sit well with me. . .)

David Abrahams wrote:
> "Andrew Koenig" <ark at acm.org> writes:
> 
> 
>>I happened to look at PEP 276 and was struck by the thought that PEP 276 is
>>really an implementation of the well-known set-theoretic construction of the
>>natural numbers that defines each natural number as the set of all smaller
>>ones.  In other words, it defines 0 as the empty set, 1 as the set whose
>>only element is 0, 2 as the set whose elements are 0 and 1, and so on.
>>
>>Indeed, under PEP 276, sets.Set(42) would be a set with 42 elements that
>>would be exactly the canonical representation of 42.
> 
> 
> ?? Simple Iterator For Ints ?? 
> http://www.python.org/peps/pep-0276.html
> Are you sure you mean 276?
> 

I think he does.

Although I think to get the set Andrew wanted with an int type sporting
an __iter__ method, you would need to write sets.Set(*42), rather than
just writing '42' on its own.

Which actually suggests strongly to me that if PEP-276 were to be
implemented, it should be done by special casing ints in for-loop
generation rather than by having ints support the iterator protocol. Far
  too much potential for cryptic usage of the latter approach!

And you could still get the set-theoretic representation using a
generator expression: sets.Set(x for x in 42)

Cheers.
Nick.

-- 
Nick Coghlan               |     Brisbane, Australia
Email: ncoghlan at email.com  | Mobile: +61 409 573 268