My PR from yesterday #6302 is pretty much the same. It failed CI twice on the Azure py37 job, timing out on the testing/test_assertion.py::test_exception_handling_no_traceback test. It seems like this one is going down the same path. I haven't looked but not sure how this affects this test, and it doesn't reproduce locally.

Oh really sorry @bluetech, I've completely missed the notification on #6302 (in fact I still can't see it in my Inbox, weird), otherwise I wouldn't have opened this one. 😓

Sorry if already discusses somewhere, the long term plan is to remove the _pytest package in favor of this one?

Sorry if already discusses somewhere, the long term plan is to remove the _pytest package in favor of this one?

No, the pytest package contains only symbols of the public API, while _pytest contains the actual codebase.

It is incredible how it is basically impossible to restart an Azure job... Clicking on Rebuild on the menu below does absolutely nothing:

the failure is quite weird, I can't reproduce it w/ 3.7.5x64 locally running tox -e py37-twisted-numpy 🤔

the failure is quite weird, I can't reproduce it w/ 3.7.5x64 locally running

Me neither... just pushed a skip for that test to see if it is that test the culprit, or something else.

Weird, it is just that test who seems to be stucking CI... 🤔

Rá looking at the test things are clearer: the test uses multiprocessing, so having the __main__ guard as @RonnyPfannschmidt pointed out is mandatory. 😁