When (and why) to use del?

[rdmurray at bitdance.com]

On Tue, 9 Dec 2008 at 18:55, Duncan Booth wrote:
> Albert Hopkins <marduk at letterboxes.org> wrote:
>
>>         def otherfunction():
>>             try:
>>                 # some stuff
>>             except SomeException, e:
>>                 # more stuff
>>                 del e
>>             return
>>
>>
>> I think this looks ugly, but also does it not hurt performance by
>> preempting the gc?  My feeling is that this is a misuse of 'del'. Am I
>> wrong?  Is there any advantage of doing the above?
>>
> It is probably a complete waste of time, but there are situations where
> code similar to this can be useful:
>
>        def otherfunction():
>            try:
>                # some stuff
>            except SomeException, e:
>                # more stuff
>                del e
>                raise
>            return
>
> The point of code like this is that when a function exits by throwing an
> exception the traceback includes a reference to the stack frame and all the
> local variables. In some situations that can result in large data
> structures not being freed for a very long time (e.g. until another
> exception is thrown that is handled at the same level). So, *in very
> specialised situations* it may be important to delete particular names from
> the local namespace.

If I'm reading http://www.python.org/dev/peps/pep-3110/ right, Python
3.0 eliminates even this use case :)

I have had occasions to use del, when I don't want the variable in the
namespace anymore for some reason (for example, when I'm later going to
be scanning the namespace for some reason).

--RDM