Knowledge baseExam Center
ENEnglish
Choose language13

CKTIS / Python

functools.cache

Memoizes function results in an unbounded cache keyed by call arguments.

Version: Python 3.14; repository image, production runtime not attested

Syntax

@functools.cache

Memoizes function results in an unbounded cache keyed by call arguments.

A repeated call with the same cache key reuses the stored result. An unbounded cache can retain arguments indefinitely.

An unbounded memoizer equivalent to lru_cache(maxsize=None) stores results by a hashable call key. Distinct argument forms, including keyword order, may create separate keys. Arguments and results remain referenced until the cache is cleared or discarded. Concurrent misses can compute a value twice. Both square(4) calls return 16, but calls records only one 4. In CPython, f(1) and f(1.0) produce distinct keys despite 1 == 1.0.

from functools import cache
@cache
def f(x):
    return object()
a = f(4)
assert f(4) is a and f.cache_info().hits == 1
f.cache_clear()
assert f(4) is not a
try:
    f([])
except TypeError:
    pass
else:
    raise AssertionError
f.cache_clear()
assert f(1) is not f(1.0) and f.cache_info().misses == 2

Examples

import functools
calls = []
@functools.cache
def square(n):
    calls.append(n)
    return n * n

print(square(4), square(4), calls)

Result

16 16 [4]

An unbounded memoizer equivalent to lru_cache(maxsize=None) stores results by a hashable call key. Distinct argument forms, including keyword order, may create separate keys. Arguments and results remain referenced until the cache is cleared or discarded. Concurrent misses can compute a value twice. Both square(4) calls return 16, but calls records only one 4. In CPython, f(1) and f(1.0) produce distinct keys despite 1 == 1.0.

Sources