René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Add tests to prevent regressions with the combination of `ctypes` and metaclasses.","articleBody":"# Feature or enhancement\n\n### Proposal:\n\nThere were [breaking changes to `ctypes` in Python 3.13](https://docs.python.org/3.13/whatsnew/3.13.html#ctypes).  \nProjects like [`comtypes`](https://github.com/enthought/comtypes/) and [`pyglet`](https://github.com/pyglet/pyglet), which implement functionality by combining `ctypes` and metaclasses, no longer work unless their codebases are updated.\nWe discussed what changes such projects need to make to their codebases in gh-124520.\n\nI believe the easiest and most effective way to prevent regressions from happening again is to add tests on the `cpython` side.\n\nI wrote a simple test like the one below.\n\n```py\n# Lib/test/test_ctypes/test_c_simple_type_meta.py?\nimport unittest\nimport ctypes\nfrom ctypes import POINTER, c_void_p\n\nfrom ._support import PyCSimpleType, _CData, _SimpleCData\n\n\nclass PyCSimpleTypeAsMetaclassTest(unittest.TestCase):\n    def tearDown(self):\n        # to not leak references, we must clean _pointer_type_cache\n        ctypes._reset_cache()\n\n    def test_early_return_in_dunder_new_1(self):\n        # this implementation is used in `IUnknown` of comtypes.\n\n        class _ct_meta(type):\n            def __new__(cls, name, bases, namespace):\n                self = super().__new__(cls, name, bases, namespace)\n                if bases == (c_void_p,):\n                    return self\n                if issubclass(self, _PtrBase):\n                    return self\n                if bases == (object,):\n                    _ptr_bases = (self, _PtrBase)\n                else:\n                    _ptr_bases = (self, POINTER(bases[0]))\n                p = _p_meta(f\"POINTER({self.__name__})\", _ptr_bases, {})\n                ctypes._pointer_type_cache[self] = p\n                return self\n\n        class _p_meta(PyCSimpleType, _ct_meta):\n            pass\n\n        class _PtrBase(c_void_p, metaclass=_p_meta):\n            pass\n\n        class _CtBase(object, metaclass=_ct_meta):\n            pass\n\n        class _Sub(_CtBase):\n            pass\n\n        class _Sub2(_Sub):\n            pass\n\n\n    def test_early_return_in_dunder_new_2(self):\n        # this implementation is used in `CoClass` of comtypes.\n\n        class _ct_meta(type):\n            def __new__(cls, name, bases, namespace):\n                self = super().__new__(cls, name, bases, namespace)\n                if isinstance(self, _p_meta):\n                    return self\n                p = _p_meta(\n                    f\"POINTER({self.__name__})\", (self, c_void_p), {}\n                )\n                ctypes._pointer_type_cache[self] = p\n                return self\n\n        class _p_meta(PyCSimpleType, _ct_meta):\n            pass\n\n        class _Base(object):\n            pass\n\n        class _Sub(_Base, metaclass=_ct_meta):\n            pass\n\n        class _Sub2(_Sub):\n            pass\n\n```\n\nIf no errors occur when this test is executed, we can assume that no regressions have occurred with the combination of `ctypes` and metaclasses.\nHowever, I feel that this may not be enough. What should we specify as the target for the `assert`?\n\n### Has this already been discussed elsewhere?\n\nNo response given\n\n### Links to previous discussion of this feature:\n\nhttps://github.com/python/cpython/issues/124520\n\n\u003c!-- gh-linked-prs --\u003e\n### Linked PRs\n* gh-125881\n* gh-125987\n* gh-125988\n* gh-126126\n* gh-126275\n* gh-126276\n\u003c!-- /gh-linked-prs --\u003e\n","author":{"url":"https://github.com/junkmd","@type":"Person","name":"junkmd"},"datePublished":"2024-10-21T11:33:32.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":7},"url":"https://github.com/125783/cpython/issues/125783"}