Re: how to "free" an object/var ?

On Jan 31, 7:34 am, Steven D'Aprano <s...@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
wrote:
On Tue, 30 Jan 2007 23:22:52 -0800, Paddy wrote:
As far as I know there is no way to force the deletion of an object
even if it is in use. This is a Good Thing.

--
Steven D'Aprano

The folowing will make the data available for garbage collection no
matter what references it:

l = ["data"] *10
l
['data', 'data', 'data', 'data', 'data', 'data', 'data', 'data', 'data',
'data']
l2 = l
l[:] = []
l2
[]

Sort of.

What happens is that both l and l2 are names referring to the same list.
After executing l[:] the *contents* of the list change, from lots of
"data" to nothing. Naturally both l and l2 see the change, because they
both point to the same list. But the original contents of the list still
exist until the garbage collector sees that they are no longer in use,
and then frees them.

You can prove this by doing something like this:

data = "\0"*1000000 # one (decimal) megabyte of data
L = [data] # put it in a list
L[:] = [] # "free" the contents of the list
assert len(data) == 1000000
# but the megabyte of data still there

Again, you can't force Python to free up memory that is still in use.
If you could, that would be a bug.

--
Steven D'Aprano

Thanks Stephen for explaining my answer a bit more.

The [:] = [] trick should be of most use when you go on to allocate
large amounts of data in your program once again, when the gc coud
then make some of the same memory available for the new objects.
As others have said, getting Python to give back memory to the
underlying OS is murder - The best way I have of doing that is to
split your program into two and chain their execution, which allows
the OS to reclaim the memory used by the first Python process when it
finishes.

- Paddy.

.

Relevant Pages

  • Re: Writing huge Sets() to disk
    ... > references in python. ... > state and taking say 600MB, pushes it's internal dictionaries ... Almost anything you do copies references. ... that this code takes a lot of extra memory. ...
    (comp.lang.python)
  • Re: 2.6, 3.0, and truly independent intepreters
    ... just the GIL being in place, but of course it's there for a reason. ... Python faster on single core machines and more stable on multi core ... Other language designers think the same way. ... with languages that use memory pointers, have the potential to get out ...
    (comp.lang.python)
  • Re: Finding the instance reference of an object
    ... "variable" is commonly thought of as a fixed location in memory ... dinosaurs and instead compare Python to any modern OOP language. ... variables are just names/aliases for *references* to ...
    (comp.lang.python)
  • Re: Finding the instance reference of an object
    ... "variable" is commonly thought of as a fixed location in memory ... dinosaurs and instead compare Python to any modern OOP language. ... Is there any way to prove that, without delving into the Python ... variables are just names/aliases for *references* to ...
    (comp.lang.python)
  • Re: why cannot assign to function call
    ... a porch the same if all its ... actual Python objects tend to be mutable only if they are ... In CPython, the id is given by the memory location of the object, which ... significant changes and replacements and add-ons that nevertheless don't ...
    (comp.lang.python)