Python resolves the scope of the parameters of a function in two ways:
- When the function is defined
- When the function is called
When the function is defined Consider this sample program which has a function adder(a, b) which adds an element a to the list b and returns the list b. The default value of b is [1, 8, 11]. Code :
Python3
def adder(a, b = [1, 8, 11]): b.append(a) return bprint(adder(1))print(adder(2))print(adder(9))print(adder(2, [3, 6])) |
Expected Output : [1, 8, 11, 1] [1, 8, 11, 2] [1, 8, 11, 9] [3, 6, 2]
Actual Output : [1, 8, 11, 1] [1, 8, 11, 1, 2] [1, 8, 11, 1, 2, 9] [3, 6, 2]
Time complexity: O(1)
Auxiliary space: O(1)
Observe that even though we don’t pass a parameter in the 2nd and 3rd call, the default value of b no longer remains [1, 8, 11]. It is because of the fact that Python resolves the reference to name of a default parameter in the scope of a function only once, when it is defined and not every time it is called. Even if a and b were to be any other mutable type like dict and set, the program would have worked in the same way. What do we do if we want the default value of b to always be [1, 8, 11] ? Inside the function, we can check if b was passed while calling the function or not. If it was not, then we reassign the value of b to [1, 8, 11]. When the function is called Now, consider this program which tries to store lambda function objects (to calculate the square of numbers between 0 and 4) inside a list and call them one by one. Code :
Python3
l = []for i in range(5): l.append(lambda : i**2)for j in range(5): print(l[j]()) |
Expected Output : 0 1 4 9 16
Actual Output : 16 16 16 16 16
However, the result is not what was expected from the program. It is because Python resolves the reference to names of non-default parameters in the scope of a function when it is called and not when it is defined. Here the parameter is i. When the loop terminates, the value of i is 4 and when we call any of the lambda functions stored in l, 4**2 = 16 is returned. What do we do to make the program work the way we expected? Just make i a default parameter for the lambda function and let python’s scope resolution behaviour do that for you.
