Why list comprehensions create a function internally?

This is disassembly of a list comprehension in python-3.10:

Python 3.10.12 (main, Jun 11 2023, 05:26:28) [GCC 11.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import dis
>>> 
>>> dis.dis("[True for _ in ()]")
  1           0 LOAD_CONST               0 (<code object <listcomp> at 0x7fea68e0dc60, file "<dis>", line 1>)
              2 LOAD_CONST               1 ('<listcomp>')
              4 MAKE_FUNCTION            0
              6 LOAD_CONST               2 (())
              8 GET_ITER
             10 CALL_FUNCTION            1
             12 RETURN_VALUE

Disassembly of <code object <listcomp> at 0x7fea68e0dc60, file "<dis>", line 1>:
  1           0 BUILD_LIST               0
              2 LOAD_FAST                0 (.0)
        >>    4 FOR_ITER                 4 (to 14)
              6 STORE_FAST               1 (_)
              8 LOAD_CONST               0 (True)
             10 LIST_APPEND              2
             12 JUMP_ABSOLUTE            2 (to 4)
        >>   14 RETURN_VALUE

From what I understand it creates a code object called listcomp which does the actual iteration and return the result list, and immediately call it. I can't figure out the need to create a separate function to execute this job. Is this kind of an optimization trick?

The main logic of creating a function is to isolate the comprehension’s iteration variable^{peps.python.org}.

By creating a function:

Comprehension iteration variables remain isolated and don’t overwrite a variable of the same name in the outer scope, nor are they visible after the comprehension

However, this is inefficient at runtime. Due to this reason, python-3.12 implemented an optimization called comprehension inlining(PEP 709)^{peps.python.org} which will no longer create a separate code object^{peps.python.org}.

Dictionary, list, and set comprehensions are now inlined, rather than creating a new single-use function object for each execution of the comprehension. This speeds up execution of a comprehension by up to two times. See PEP 709 for further details.

Here is the output for the same code disassembled with python-3.12:

>>> import dis
>>> dis.dis("[True for _ in ()]")
  0           0 RESUME                   0

  1           2 LOAD_CONST               0 (())
              4 GET_ITER
              6 LOAD_FAST_AND_CLEAR      0 (_)
              8 SWAP                     2
             10 BUILD_LIST               0
             12 SWAP                     2
        >>   14 FOR_ITER                 4 (to 26)
             18 STORE_FAST               0 (_)
             20 LOAD_CONST               1 (True)
             22 LIST_APPEND              2
             24 JUMP_BACKWARD            6 (to 14)
        >>   26 END_FOR
             28 SWAP                     2
             30 STORE_FAST               0 (_)
             32 RETURN_VALUE
        >>   34 SWAP                     2
             36 POP_TOP
             38 SWAP                     2
             40 STORE_FAST               0 (_)
             42 RERAISE                  0
ExceptionTable:
  10 to 26 -> 34 [2]

As you can see, there is no longer a MAKE_FUNCTION opcode nor a separate code object. Instead python-3.12 uses LOAD_FAST_AND_CLEAR^{docs.python.org}(at offset 6) and STORE_FAST(at offset 30) opcodes to provide the isolation for the iteration variable.

Quoting from the Specification section^{peps.python.org} of the PEP 709:

Isolation of the x iteration variable is achieved by the combination of the new LOAD_FAST_AND_CLEAR opcode at offset 6, which saves any outer value of x on the stack before running the comprehension, and 30 STORE_FAST, which restores the outer value of x (if any) after running the comprehension.

In addition to that, in python-3.12 there is no longer a separate frame for the comprehension in tracebacks.

>>> [1 / 0 for i in range(10)]
Traceback (most recent call last):
 File "<stdin>", line 1, in <module>
 File "<stdin>", line 1, in <listcomp>
ZeroDivisionError: division by zero

>>> [1 / 0 for i in range(10)]
Traceback (most recent call last):
 File "<stdin>", line 1, in <module>
ZeroDivisionError: division by zero

And here is the benchmark results^{peps.python.org}(measured with MacOS M2):

$ python3.10 -m pyperf timeit -s 'l = [1]' '[x for x in l]'
Mean +- std dev: 108 ns +- 3 ns
$ python3.12 -m pyperf timeit -s 'l = [1]' '[x for x in l]'
Mean +- std dev: 60.9 ns +- 0.3 ns