When do you need to use self in a nested function?

Question

Why do we need to use self for ans, but not min1 in the following code (via LeetCode)?

def findSecondMinimumValue(self, root):
    self.ans = float('inf')
    min1 = root.val

    def dfs(node):
        if node:
            if min1 < node.val < self.ans:
                self.ans = node.val
            elif node.val == min1:
                dfs(node.left)
                dfs(node.right)

    dfs(root)
    return self.ans if self.ans < float('inf') else -1

`min1` is a local variable and `ans` is an attribute of `self` because you created them that way. — khelwood, Jul 07 '20 at 22:32
@khelwood This is a solution from leetcode and my question is why we need self for ans but not for min1? If I take away self from ans the code will not run correctly. — jlin90, Jul 07 '20 at 22:43
@AMC Oh, haha! I see what you mean though: the class definition is missing. — wjandrea, Jul 07 '20 at 22:55

raccoons · Answer 1 · 2020-07-07T23:05:01.880

1

min1 is local to the function body. Its name only exists within the body of findSecondMinimumValue(). Once your code exits that function, min1 won't be available for use.

ans is an attribute that belongs to the class instance and has to be accessed through self. If you created two different instances of whatever class this method is a part of, that would mean each instance has its own version of self.ans, self being a name that refers to the instance that calls the method and self.ans therefore referring to an instance's ans attribute.

As an example:

class Example:
    foo = "CLASS attribute"

    def __init__(self):
        self.bar = "INSTANCE attribute"

a = Example()
b = Example()
a.bar = "INSTANCE attribute changed"
print(a.bar) # -> "INSTANCE attribute changed"
print(b.bar) # -> "INSTANCE attribute" because changing a.bar doesn't affect b.bar

Example.foo = "CLASS attribute changed"
print(a.foo == b.foo) # -> True

edited Jul 07 '20 at 23:05

answered Jul 07 '20 at 22:30

raccoons

420
3
16

1

You're very close, but `ans` is an *attribute* of the instance, not a variable. Instances don't have variables per se. – wjandrea Jul 07 '20 at 22:49
Gotcha, wasn't sure about the exact naming of it. Corrected the post. – raccoons Jul 07 '20 at 22:51
1

Better, but it's still a bit off. Attributes don't have scope; they're attached to an object. – wjandrea Jul 07 '20 at 22:52

score 0 · Answer 2 · answered Jul 07 '20 at 22:52

I guess is much simpler. You're recursively changing self.ans. That's why, if you don't use self, you have to define and pass an extra variable, I guess.

Just another version, which'd pass through:

class Solution:
    def findSecondMinimumValue(self, root):
        self.second = float('inf')

        def traverse(node):
            if not node:
                return

            if root.val < node.val < self.second:
                self.second = node.val

            traverse(node.left)
            traverse(node.right)

        traverse(root)
        return -1 if self.second == float('inf') else self.second

References

For additional details, you can see the Discussion Board. There are plenty of accepted solutions with a variety of languages and explanations, efficient algorithms, as well as asymptotic time/space complexity analysis^{1, 2} in there.

When do you need to use self in a nested function?

2 Answers2

References