What is a closure, and what is the late-binding problem inside loops?

Short answer

A closure is an inner function that keeps a reference to variables of the enclosing scope, but the variable's value is read at call time, not creation time, so all functions created in a loop see the loop variable's last value.

Explanation

A closure refers to the variable itself (a cell), not to a copy of its value. When you create lambdas inside a loop and run them later, they all read the final value of the loop variable. Solutions: bind the value as a default parameter (lambda n=n: n), use functools.partial, or use a factory function that creates a new scope for each value. To modify an enclosing variable from the inner function, use nonlocal.

Code example

callbacks = [lambda: n for n in range(3)]
print([f() for f in callbacks])  # [2, 2, 2]

callbacks = [lambda n=n: n for n in range(3)]
print([f() for f in callbacks])  # [0, 1, 2]

def counter():
    count = 0
    def inc():
        nonlocal count
        count += 1
        return count
    return inc

Common mistakes

• Creating callbacks inside a loop and expecting each to carry its own value.
• Forgetting nonlocal when modifying an enclosing variable, which raises UnboundLocalError.

Follow-up questions

• What is functools.partial for?

Sources

• Python FAQ — Why do lambdas defined in a loop all return the same result?