run ci against gevent/greenlet monkeypatching

[RonnyPfannschmidt]

#3503 demonstrated that testing against stdlib-patchers provides a safety net for users

[pytestbot]

GitMate.io thinks possibly related issues are #3423 ([Feature request] Colorized output when run in CI), #2544 (Run tests truly programmatically), #481 (Running doctests with -s fails?), #1985 (Monkeypatching "compile" breaks pytest sometimes), and #1484 (running py.test).

[nicoddemus]

Gave this a quick shot on master by creating testing/conftest.py:

from gevent import monkey
monkey.patch_all()

Unfortunately that doesn't break the tests. Any ideas?

[RonnyPfannschmidt]

@nicoddemus isn't master already fixed, prior to your patch it should fail

this is mainly intended to catch minor issues

[nicoddemus]

I tried the above without the fix for gevent.

[RonnyPfannschmidt]

oh, extra fun then

[RonnyPfannschmidt]

or perhpas its precisely because we patched too early

what happens if the monkeypatch happens after pytest is initialized but before tests are executed

[nicoddemus]

what happens if the monkeypatch happens after pytest is initialized but before tests are executed

When the conftest is imported pytest has been initialized already, so that's been tried. I also tried to patch even early by using a wrapper script:

from gevent import monkey
monkey.patch_all()
import pytest
pytest.main()

But still no good. 😕

[RonnyPfannschmidt]

i see, interesting

[RonnyPfannschmidt]

closing as not actionable - if the next one comes up we can try again, this one was on the bench for long enough