Nathan Goldbaum

Closing since we aren't planning to support Python 3.13 on the 2.0 branch.

> I was also worried about warnings, but it seems that catch_warnings is handled in Python 3.14, so pytest's warnings plugin should work at least. And in-practice I haven't seen...

> context aware capture for stdio/logging with the awareness that posix level capture is never thread-safe (this might also help to resolve some long standing developer experience bugs as people...

Liam's summary seems like an excellent plan and would substantially help ecosystem efforts to improve multithreaded test coverage. Currently there's lots of manual patching that's needed to work around pytest...

> To limit the scope of this, I think we should limit the usage of --threads (just gonna assume that will be the flag 😀 ) to the freethreaded build...

It'll require upstream support in PyO3 and maybe CFFI first. Once one of PEP 803 or 809 is accepted and there's a new abi to use in 3.15, I'll be...

I think this is happening because `sys.modules` can change "underneath" a thread - it's even documented to do this: https://docs.python.org/3/library/sys.html#sys.modules, and CPython directly using `sys.modules` in `pkgutil.walk_packages` is unsafe for...

@davidhewitt would you like a hand getting this rebased and updated for 0.28?

@davidhewitt FYI - I just force-pushed a rebase to this branch. The only nontrivial bit was the fix for the conflict with https://github.com/PyO3/pyo3/commit/85e6507f261e8a288fc699cb25bd6fd6986d2101#diff-f97e0150474170093c7508d4ac482d2ab24d5325468b860c85aa4da13024917a, but I think I have it right...

I created a C backport for `IsUniquelyReferenced`, see: https://github.com/python/pythoncapi-compat/pull/149 That might be useful for inspiration if anyone wants to write a safe wrapper that works on older Python versions.