After reworking the patch to backport the pre-initialization embedding tests to 3.7.0a1, they *both* failed, as that was before Victor fixed Py_DecodeLocale to work prior to initialization again.

As a result, I tried https://github.com/python/cpython/commit/43605e6bfa8d49612df4a38460d063d6ba781906 in November, after the initial merge of the PEP 432 working branch, but just before the start of Victor's follow on refactoring to really take advantage of the new internal structures: double failure there as well.

https://github.com/python/cpython/commit/f5ea83f4864232fecc042ff0d1c2401807b19280 is the next point I checked (just before Eric's September changes to global state management), and it looks like we have a new chief suspect, as applying the attached "earlier-tree" version of the patch at this point in the history gives two passing test cases :)

Next I'll work out a git bisect run based on those two start and end commits and the "earlier-tree" patch, and attempt to narrow the failure down to a specific commit (probably tomorrow).


(Even without that bisect run though, the warnoptions changes in https://github.com/python/cpython/commit/2ebc5ce42a8a9e047e790aefbf9a94811569b2b6#diff-f38879f4833a6b6847e556b9a07bf4ed are now looking *very* suspicious)