test_completion_param_annotations fails in Python 3.13

[carlwgeorge]

Fedora has switched to Python 3.13 in its Rawhide branch in preparation for Fedora 41. This has brought to our attention that jedi's test_completion_param_annotations test fails with this version of Python.

🐍 .py3.13 ❯ pytest -k test_completion_param_annotations
========================================== test session starts ===========================================
platform linux -- Python 3.13.0b2, pytest-6.2.5, py-1.11.0, pluggy-1.5.0
rootdir: /home/carl/development/jedi, configfile: pytest.ini, testpaths: jedi, test
collected 3888 items / 3887 deselected / 1 selected                                                      

test/test_api/test_interpreter.py F                                                                [100%]

================================================ FAILURES ================================================
___________________________________ test_completion_param_annotations ____________________________________

    def test_completion_param_annotations():
        # Need to define this function not directly in Python. Otherwise Jedi is too
        # clever and uses the Python code instead of the signature object.
        code = 'def foo(a: 1, b: str, c: int = 1.0) -> bytes: pass'
        exec(code, locals())
        script = jedi.Interpreter('foo', [locals()])
>       c, = script.complete()
E       ValueError: not enough values to unpack (expected 1, got 0)

test/test_api/test_interpreter.py:315: ValueError
======================================== short test summary info =========================================
FAILED test/test_api/test_interpreter.py::test_completion_param_annotations - ValueError: not enough va...
=================================== 1 failed, 3887 deselected in 0.55s ===================================

[PeterJCLaw]

@carlwgeorge thanks for reporting this. I've been working on Python 3.13 support and while the majority of the updates have been in tests, there have been some source changes to ensure correct behaviour.

I wanted to let you know that we're working on this and while I'm hopeful of getting support ready in the next few weeks I can't commit to a timeline. I appreciate from the link you shared to the Fedora wiki that that might be tight for your own timelines, so thought it best to let you know where we stand at the moment.

[carlwgeorge]

Thanks for the update. No worries about the timeline, it's quite common that things don't line up ideally between projects. I certainly don't want you to feel like myself or the Fedora Project is rushing you. For now what I think we'll do in the Fedora package is skip that test just so that we can get a Python 3.13 build completed, even if it isn't perfect. If there are any commits you feel are really important for Python 3.13 compatibility, even if not final or included in a tagged release yet, I can include those as patches in the package. I can even include all of #2003 as a patch if you would like. We have several options here. We can always refresh the patch or even update to a new version after the Fedora 41 release (it isn't frozen like Debian).

[carlwgeorge]

Closing the loop here, since #2003 was merged I went ahead and included it as a patch to the Fedora package. That let us complete a Python 3.13 build, which will unblock the installation of several other packages. Thanks!

[PeterJCLaw]

For info: you may also want to pull in #2008 too.