Eric Snow

That step 4 is mostly obsolete. With the solution I merged for the datetime C-API, the problems of isolation for subinterpreters there are mostly resolved. We only have to make...

> I think `PyDate*_Check(op)` API macros are allowed to successfully compare a C-API type with a normal `_datetime` (heap) type. That shouldn't be a problem. The fact that the types...

Here's the updated TODO for beta 2: 1. basic multi-phase init ([gh-119373](https://github.com/python/cpython/pull/119373)) 2. convert to heap types (except for 5 types exposed by datetime C-API) 3. convert to module state...

@erlend-aasland, the backport of that multiphase PR broke a bunch of the 3.13 buildbots, so I'm reverting it. We can regroup and sort that out.

For example: https://buildbot.python.org/all/#builders/1508/builds/69 ``` test_utc_alias (test.datetimetester.TestModule_Pure.test_utc_alias) ... ok test_utc_alias (test.datetimetester.TestModule_Fast.test_utc_alias) ... FAIL ====================================================================== FAIL: test_utc_alias (test.datetimetester.TestModule_Fast.test_utc_alias) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/buildbot/buildarea/3.13.cstratak-RHEL7-x86_64/build/Lib/test/datetimetester.py", line 98, in test_utc_alias self.assertIs(UTC, timezone.utc)...

This may be a transient failure but it's probably still worth checking.

Good question. If anything changed in that regard, it wouldn't be specific to this buildbot (that I'm aware of). Also, I took a quick look at the recent commit log...

We could either skip CPU-intensive tests on the buildbot or increase the buildbot's timeout.

(set "timeout_factor" for the worker in master.cfg in the buildmaster-config repo)

Does applying `@suppress_immortalization()` make a difference. I'm not suggesting that as a solution (there's probably a deeper issue to solve), but whether or not the decorator makes a difference might...