…alization as in the plugin.

The plugin code creates a new event loop from the current event loop policy, whereas the fixture in the Hypothesis test uses get_event_loop. In Python 3.10 the use of get_event_loop was deprecated and causes the Hypothesis tests to fail.

Signed-off-by: Michael Seifert <[email protected]>

…has no effect.

Under Python 3.10, the remove_loop fixture raises:
"DeprecationWarning: There is no current event loop"
This means that the remove_loop fixture effectively does nothing.

The test case is still kept, because it seemingly aims to test that the "asyncio" marker works inside test classes when no "event_loop" fixture is explicitly specified.

Signed-off-by: Michael Seifert <[email protected]>

…the asyncio marker in class-based tests.

Rephrased test names and comments to make the test intention more clear.

Signed-off-by: Michael Seifert <[email protected]>

…osed.

Under Python 3.10, the tests raise:
"DeprecationWarning: There is no current event loop"
This means that the asyncio.get_event_loop does not return any existing loop. It creates a new loop as a side effect instead. Therefore, test_1 and test_3 will always be successful.

However, it would be wrong to simply delete the tests without a replacement. Although, we cannot retrieve the default loop with get_event_loop, we can still retrieve it from the event loop policy. The new tests do exactly that and assert that the loop is a different object than in the tests before.

Signed-off-by: Michael Seifert <[email protected]>

…ecationWarning on Python 3.10.

The code was adjusted to use asyncio.get_event_loop_policy().get_event_loop() rather than the deprecated asyncio.get_event_loop().
Some additional comments were added to clarify the circumstances under which the test can fail.

Signed-off-by: Michael Seifert <[email protected]>

Signed-off-by: Michael Seifert <[email protected]>

Signed-off-by: Michael Seifert <[email protected]>

Adjusted changelog and setup.py accordingly.

Signed-off-by: Michael Seifert <[email protected]>