There seems to be quite some code duplication between the two plugins - do they really need to be separate?

The certainly look similar, but the similarity is just structural, not logical. IMO it would be wrong to couple them -- they couldn't be disabled separately, and I suspect will cause maintainability issues if new requirements ever come up, a process which is nicely described in this article and is pretty accurate from my experience.

Thanks for the work on this, looking forward to making use of it! Does anyone know which pytest release this is likely to be released in?

It will go in 6.2, which we plan to release this Saturday! 🎉

#8101

Could someone explain what this new feature means? I read the document of PytestUnraisableExceptionWarning but still have no idea what it means:

Unraisable exceptions are exceptions raised in __del__ implementations and similar situations when the exception cannot be raised as normal.

Thanks!

Here is an example from the tests:

        class BrokenDel:
            def __del__(self) -> None:
                raise ValueError("del is broken")

        def test_it() -> None:
            obj = BrokenDel()
            del obj

Previously the ValueError would go unnoticed, since the test does pass and the output is captured. Now pytest issues a PytestUnraisableExceptionWarning warning for this test, so you can better notice the ValueError problem.

Is there a way to propagate this warning to an error to let the tests fail?

@flyser see https://docs.pytest.org/en/stable/warnings.html#warnings-capture

you can set:

# pyproject.toml
[tool.pytest.ini_options]
filterwarnings = ["error"]

I like to use the most strict config I can:

# pyproject.toml
[tool.pytest.ini_options]
addopts = ["--strict-config", "--strict-markers"]
filterwarnings = ["error"]
xfail_strict = true