freethreaded support

[15r10nk]

solves #192

[davidhewitt]

Thanks for working on this! If you have specific things you'd like help on, myself or @ngoldbaum may be of use :)

[15r10nk]

I think the two remaining problems are the following:

• a new way to format the code with black (see Black's public API psf/black#779 (comment))
• pydantic support for 3.13t (needed to test the pydantic support)

[davidhewitt]

I guess for black, you might be able to just put a lock around that? Sadly warnings.catch_warnings is also not thread safe (I think there might be new APIs considered for 3.14).

Pydantic support is on the way, follow pydantic/pydantic#10483 - probably will be ready in ~1 month.

[15r10nk]

I guess for black, you might be able to just put a lock around that?

This does not work. The problem is that another thread is writing something to sys.stdout at the same time the CliRunner is mocking it to capture the output of black.
I will have to use the internal api functions here. I just hope that I don't have to re-implement the config option parsing of black.

The following code shows the problem with stdout.

# /// script
# dependencies = [
#   "pytest",
#   "pytest-run-parallel"
# ]
# ///


from contextlib import redirect_stdout
from io import StringIO
import time


def test_redirect():
    text=StringIO()
    with redirect_stdout(text):
        print("hello")
        time.sleep(1)
        print("hello")

    assert text.getvalue()=="hello\nhello\n"


if __name__ == "__main__":
    import pytest
    pytest.main(["--parallel-threads=5",__file__])

contextlib.redirect_stdout is also not thread save for the same reason. It is modifying global state. I don't know if stdin/stdout can be fixed in the same way.

[ngoldbaum]

Maybe open a CPython bug? It seems reasonable to me for python to add locking for global state in the C standard library.

[15r10nk]

I created an issue python/cpython#130148