I am back in school, to finally get my Bachelor's degree. This started as a writing assignment on void functions for my Programming fundamentals class, but I couldn't let it stop there.
If a Python function does not return a result, the output from a function is None and that function is a void function. Print is an example of a voidfunction. Although it adds text to the standard output, that printed value isn’t stored when assigned to a value:
>>> r = print('r')
'r'
>>> r
>>>
Most Python data types are immutable, meaning their value cannot be modified without changing the object reference. See Python official documentationfor more info on Python's built-in types.As a high level language, Python is taking care of our objects behind the scenes. We can continue to modify the same variable name, but the ID of the object changes when Python creates another object. This immutability has an effect of protecting a local value within afunction from modifying the value of a variable outside of a function. In the following example, x references an integer that is modified with math, causing x to be recreated with a different value and ID:
>>> x = 10
>>> id(x)
2428003641936
>>> x += 1
>>> id(x)
2428003641968
We can pass x as an argument that is modified within the function, without modifyingthe value of x outside of the function:
>>> x = 2
>>> def add_two(y):
... y += 2
... return y
...
>>> add_two(x)
4
>>> x
2
However, dictionaries and lists are mutable. Notice how the ID is the same after a modification:
>>> r = [1]
>>> id(r)
2428004605312
>>> r += [3]
>>> r
[1, 3]
>>> id(r)
2428004605312
This leads to a very interesting possibility of a void function with a side effect. Side effects make a change to something other than what it’s output is assigned to. The following shows the same mutability for a dictionary and then an example of a void function with a side effect of changing the dictionary:
>>> r = {}
>>> id(r)
2428004605824
>>> r['name'] = 'Jessie'
>>> id(r)
2428004605824
>>> def set_no_name(myDict):
... if not myDict.get('name'):
... myDict['name'] = 'stranger'
...
>>> stranger = {}
>>> set_no_name(stranger)
>>> stranger
{'name': 'stranger'}
>>> jess = {'name': 'Jess'}
>>> set_no_name(jess)
>>> jess
{'name': 'Jess'}
To add further complexity (to both our objects and understanding), the object types of lists, tuples, and dictionaries are container objects. The items within them are stored by reference (to the IDs), not value. In the following example, a tuple is an immutable type containing a mutable list and an immutable integer. When we modify the list and the integer, the value of the list inside the tuple is changed and the value of the integer is not changed.The IDs of both the mutablle list and the immutable integer in the tuple are unchanged.
>>> r = [1,2,3]
>>> s = 3
>>> t = (r, s)
>>> id(r)
2428009990336
>>> id(s)
2428003641712
>>> id(t)
2428009991424
>>> r.append(4)
>>> s+=1
>>> t
([1, 2, 3, 4], 3)
>>> id(t)
2428009991424
>>> id(t[0])
2428009990336
>>> id(t[1])
2428003641712
The JavaScript community often uses “purity” to discuss side effects. A function with a side effect is impure, versus no side effects is called pure. Impure functions can be harder to test and harder to debug if the developer doesn’t realize there is a sideeffect. Functions with side effects are useful.Personally, I am fine with using side effects if it is a void function.A code comment of returning nothing would be useful for future me or someone else to understand the code, while a function with both a return and a side effect is a little messy. My confidence in this stance is boosted by how Python core uses this strategy for appending to a list:
>>> s = [1,2,3]
>>> id(s)
2428009990336
>>> s.append(4)
>>> s
[1, 2, 3, 4]
>>> id(s)
2428009990336