Roulette wheel

[mattia]

Il Wed, 18 Mar 2009 23:31:09 -0200, Gabriel Genellina ha scritto:

> En Wed, 18 Mar 2009 18:49:19 -0200, mattia <gervaz at gmail.com> escribió:
>> Il Wed, 18 Mar 2009 13:20:14 -0700, Aahz ha scritto:
>>> In article <49c1562a$0$1115$4fafbaef at reader1.news.tin.it>, mattia
>>> <gervaz at gmail.com> wrote:
>>>>
>>>> Yeah, and I believe that we can say the same for: 1 - t = [x*2 for x
>>>> in range(10)]
>>>> 2 - t = list(x*2 for x in range(10))
>>>> or not?
>>> The latter requires generator expressions, which means it only works
>>> with Python 2.4 or higher.  Personally, I think that if the intent is
>>> to create a list you should just use a listcomp instead of list() on a
>>> genexp.
>> Ok, so list(x*2 for x in range(10)) actually means: list((x*2 for x in
>> range(10)) --> so a generator is created and then the list function is
>> called?
> 
> Exactly. The (()) were considered redundant in this case.
> 
>> Also, dealing with memory, [...] will be deleted when the reference
>> will be no longer needed and with list(...)... well, I don't know? I'm
>> new to python so sorry if this are nonsense.
> 
> I don't completely understand your question, but *any* object is
> destroyed when the last reference to it is gone (in CPython, the
> destructor is called at the very moment the reference count reaches
> zero; other implementations may behave differently).

OK, understood. Now, as a general rule, is it correct to say:
- use generator expression when I just need to iterate over the list or 
call a function that involve an iterator (e.g. sum) and get the result, 
so the list is not necessary anymore
- use list comprehensions when I actually have to use the list (e.g. I 
need to swap some values or I need to use sorted() etc.)
Am I right?