On Wed, Nov 16, 2011 at 9:25 AM, Steven D'Aprano
<steve+comp.lang.pyt...@pearwood.info> wrote:
> Languages aren't refcounted. Or at least, *Python* isn't a refcounted
> language. CPython is a refcounted implementation. IronPython and Jython
> are not. del behaves exactly the same in IronPython and Jython as it does
> in CPython: it removes the name, which may have a side-effect of deleting
> the object.
Yes, I was sloppy there. A non-manually-memory-managed language, if
you will; it's part of Python's spec that you do NOT have to
explicitly release objects you're no longer using.
>> The del
>> statement is identical to "a = None" in terms of deleting objects;
>
> I'm not entirely sure what you arr trying to say here. I *think* you are
> trying to say is this:
>
> Given that `a` is bound to my_object initially, `del a`
> from the point of view of my_object is no different
> from re-binding a to some other object such as None
> (or any other object).
>
> which is true as far as it goes, but it fails to note that the name "a"
> no longer exists after `del a` while it does exist after `a = None`.
Right. Both actions have the same effect wrt deleting my_object; the
only connection between Python's "del" and C++'s "delete" is that,
which del shares with "a = None".
The fact is that, regardless of the Python implementation, deleting
*objects* is not the programmer's responsibility. The only thing he
can or must do is delete *names*.
del a
del globals()['a']
globals().__delitem__('a')
are all roughly equivalent (assuming that a is global). Maybe this is
the best way to explain it - that you're deleting from a "dictionary"
(which may or may not actually be implemented as a dict) of local
names.
ChrisA
--
http://mail.python.org/mailman/listinfo/python-list
