On Thu, Nov 5, 2009 at 10:39 AM, spir <denis.spir@[...].fr>  wrote:
> 
>  Thank you very much! That's exactly what I expected. Was sure my code was uselessly heavy.
 Actually, when reading the doc about iteration, I had wrongly understood that next() is req
uired, too.
> 

This is actually correct. An iterator still requires a next method.
The nice thing about generator functions is that calling them creates
a generator object, which supplies the next method for you
automatically. As below:

> >> a = PolyRange((10, 20), (30, 40))
> >> iter(a)
<generator object at 0x7fe8092c38c0> 
> >> b = iter(a)
> >> b.next()
10
> >> b.next()
11

etc. A generator expression does essentially the same thing.


>  Two additional questions (relative to things manually implemented in my original code):
>  * What about memorization of "emptyness", meaning the last item is already reached, and fo
llowing calls will all fail. This is automatic for generators, but...
>  * Then how do you restart it? With a decoupling of __iter__() and next(), it's possible to
 have both failure when empty for the same iterator (= call to next()), and
>  a new iterator returned by __iter__(), typically for a new "for" statement. Below after a 
bug correction (attributes needing initialisation):
> 
>  <snip example>
> 
>  PS: Just checked and works as expected with generator.
> 

Yes, every time you call iter(), a new generator object is created,
which works independently of other generators.

As an aside, I just thought of an even shorter implementation that
does not sacrifice clarity, using the itertools module:

#this imported at the top of the file
import itertools

def __iter__(self):
    return itertools.chain(*self.ranges)

HTH,
Hugo
_______________________________________________
Tutor maillist  -  Tutor@[...].org
To unsubscribe or change subscription options:
http://mail.python.org/mailman/listinfo/tutor