Brett Cannon <br...@python.org> added the comment:
>From Victor: * Maybe the test_importlib should before save/restore the the "Python state" rather than modifying modules You will have to be more specific than that as there is an import_state() context manager to control import state via the sys module. * Maybe module.__spec__ should leak less importlib internals: explicitly clear namespaces? Use static methods? I'm not sure how to do that. Are you saying change everything on __spec__ objects to be static methods? That won't work because the whole point of __spec__ objects is to essentially be data classes to store details of a module. * Remove module.__spec__? ... that would mean rejecting PEP 451 implemented in Python 3.4, that sounds like a major regression :-( No. You would break the world and undo years of work to clean up import semantics with no good way to go back to the old way. Can I ask why you suddenly want to throw __spec__ objects out due to a leak tied back to _weakref switching to multi-phase initialization? Also note that the use of weakrefs by importlib isn't tied to __spec__ objects but to per-module import locks in importlib._bootstrap. ---------- _______________________________________ Python tracker <rep...@bugs.python.org> <https://bugs.python.org/issue40050> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
