Is the small number cache defined in Python specification or is it an implementation detail?

Question

Python seems to have a so called "small number cache" for numbers in the range -5 to 256. We can demonstrate that with the following program:

for i in range(-7, 258):
    if id(i) == id(i + 0):
        print(i, "is cached")
    else:
        print(i, "is not cached")

The output is

-7 is not cached
-6 is not cached
-5 is cached
-4 is cached
[...]
255 is cached
256 is cached
257 is not cached

I wonder whether that's defined in the Python specification or it's an implementation detail.

@Chris_Rands I voted to re-open because I think this will make a good dupe target on it's own in the future. — juanpa.arrivillaga, Sep 11 '21 at 19:02
Might be good to include the (abbreviated) output, so we don't all have to run it ourselves. Btw `--i` also works. — no comment, Sep 11 '21 at 19:41

Thomas Weller · Accepted Answer · 2021-09-11T18:32:27.000

It's an implementation detail: chapter 3.1 of the Python language reference (all versions, from 2.7 through 3.9) says:

for immutable types, operations that compute new values may actually return a reference to any existing object with the same type and value

and

E.g., after a = 1; b = 1, a and b may or may not refer to the same object with the value one, depending on the implementation

(emphasis mine)

The current implementation keeps an array of integer objects for all integers between -5 and 256,

(emphasis mine) which is also quite clear that this might change in the future.

1 Answers1