Originally published on medium
TLDR; Avoid using an empty list on any mutable object as a default argument to a function
Working with Python for almost two years now, and still learning new things every day. Don’t think it’s something extremely advanced, don’t get me wrong sometimes it is, but not today. I wanna talk about something much more simple like default arguments.
Default arguments are the default values set on the function definition like:
def function(element=1, other_element=0):
pass
This function can be called in several ways:
- function()
- function(1000)
- function(10, 20)
The Python documentation also adds an important warning:
Important warning: The default value is evaluated only once. This makes a difference when the default is a mutable object such as a list, dictionary, or instances of most classes.
Translating this to code:
def f(a, L=[]):
L.append(a)
return L
print(f(1))
print(f(2))
print(f(3))
This will print:
[1]
[1, 2]
[1, 2, 3]
We can see that the default argument L=[] is only evaluated once, so a new list is created once when the function is defined, and the same list is used in each successive call. But you may think: “Hey, this is wrong, I want to receive a list with each individual value” like the following:
[1]
[2]
[3]
To achieve this and don’t allow the default argument to be shared between subsequent calls, you can write the function like this instead:
def f(a, L=None):
if L is None:
L = []
L.append(a)
return L
Avoid having mutable default arguments to functions, use None instead. Here is an awesome post about how to take advantage of this Python feature.