How severe does this issue affect your experience of using Ray?
- Low: It annoys or frustrates me for a moment.
I cannot debug my unit tests (using pytest) with PyCharm’s interactive debugger if any of my tests require ray.data
because attempting to import ray.data will result in a ModuleNotFoundError
.
Running the tests normally, not in debug mode, has no such problem.
My unit tests are meant to test the correctness of my actors’ main processing methods. I encountered this weird behaviour when trying to use a parquet->dataset->dataframe as my fixture data for the tests, because this is what the actors see. I also get this error inside pandas code if I try to pd.read_pickle
and the dataframe has a TensorDtype
column.
I can work around this for now by replacing any TensorDtype
columns with object types before pickling my test dataframe.