Coverage remained the same at 81.139% when pulling 9f76759 on ci-pip-pre-and-py311 into 4636fdb on master.

Something really weird is happening where the pytest gets stuck on 13%. The last test passed is

test/test_analysis/test_feature_scoring.py::FeatureScoringTestCase::test_prepare_outcomes PASSED [ 13%]

and after that it keeps running indefinitely without finishing (failing or otherwise).

In earlier testing the test directly after that

test/test_analysis/test_logistic_regression.py::LogitTestCase::test_logit FAILED [ 14%]

failed, as can be seen here in the log. The test fails because of an NumPy error as can be seen here:

>       return masked_where(~(np.isfinite(getdata(a))), a, copy=copy)
E       TypeError: ufunc 'isfinite' not supported for the input types, and the inputs could not be safely coerced to any supported types according to the casting rule ''safe''

This might be either a regression in NumPy, or expected behavirour for NumPy 1.24 (currently tested is 1.24.0rc2).

@quaquel Fixing this error might help the pytest to also continue running. Do you have any idea why this is failing and if it's expected?

without checking the test code because I don't have time for that right now, my hunch is that this is a datatype issue due to ongoing changes in the datatype system in numpy. I'll try to look at this later this week (most likely Thursday) if still necessary.

looks good and can be merged

NumPy 1.24.0 was just released, so I triggered a CI run, and now all jobs fail on the failed test case above: https://github.com/quaquel/EMAworkbench/actions/runs/3726755907

How important is this failing test? Do we need to issue a bug fix release after fixing this bug?

It at least proves the usefulness of having an early warning CI job.

It would be worth issuing a bug fix release. I'll try to look at the issue today.

Looks like it's an interaction bug between NumPy and Pandas. See issues numpy/numpy#22826 and numpy/numpy#22720, and the bugfix PR numpy/numpy#22838.

So NumPy 1.24.1 was released with this fix, and now the CI is all over the place, with builds seemingly randomly passing or getting stuck on that test. Very strange.

@quaquel Let's merge this. Even if the Ubuntu jobs only sometimes pass, it's useful to have this as an early warning.

Also, the required tests need to be updated (or removed all together for my part, we're perfectly capable of interpreting the test results ourselves).