Will a function within a function take up more RAM memory than if the 2 functions are defined separately?

user_12462568 0
Viewed 289

Say if I have the following code with a function within a function (1st method):

def a:
    # do something and local variables gets created

    def b:
        # do something and local variables gets created

and the following code where the 2 functions are defined separately (2nd method):

def a:
    # do something and local variables gets created

def b:
    # do something and local variables gets created

When I run the code, either with the 1st method or 2nd method above, which method will likely to take up more RAM memory? Or in this case, it doesn't matter and should roughly be the same?

Many thanks in advance.

ram memory memory-management function python
2 Answers

Inner functions are called closures and should not be used just to hide the function from the rest of the module. This is partially a style issue but in fact python creates a new function object for the inner function each time the outer function is called.

user_6420700

It depends on how many times you call the outer function, and on whether it returns the nested function or not.

If you have a nested function, the statement def b(...) not only assigns a new function object to the name b, but also creates a closure that encapsulates the namespace of a for b to access. This is temporary for the duration of the execution of a if a does not return b, i.e., make it non-local.

If b is a separate function object, it gets assigned to a function object just once, when the module code is run on import. It has no non-local closure namespace.

In both cases, the code of b is compiled into a code object exactly once, on module import. The function objects that are created are relatively cheap by comparison. The function objects reference the code singleton. The function objects will also always have a reference to the global namespace in both cases.

So if b is nested in a, the interpreter will create a new function object around the code of b every time a is called. It will also create a closure for the namespace of a.

If you intend to call a at least once, the non-nested version is less memory intensive since does not create closures or create multiple function objects.

Sometimes, however, design decisions are made to sacrifice some speed and memory usage for legibility, maintainability, or other considerations. You are using python after all. If, for example, b ought to stay hidden from the rest of the code, I would likely posit that the nested solution is better despite being generally more memory and processor intensive. An example is where b is an implementation detail of a, like the recursive portion of a function.

user_29887300
Related