jorisvandenbossche commented on PR #37821: URL: https://github.com/apache/arrow/pull/37821#issuecomment-1801572988
>> there is a Python failure that seems to happen consistently for the different builds / different last commits: > > That is weird - I'm trying to reproduce it here. I can't see what has changed that isn't behind a `#ifdef EMSCRIPTEN` I can reproduce it locally. When fetching this branch and building it in my existing development environment, I get the same failures. I am not sure if it's something we can fix in the test, or requires an actual code fix. The current test assumes that it gets a `KeyboardInterrupt` exception. But it seems we now get a different error, a `pyarrow.lib.ArrowCancelled`, and this isn't caught in the test: https://github.com/apache/arrow/blob/e62ec62e40b04b0bfce76d58369845d3aa96a419/python/pyarrow/tests/test_csv.py#L1418-L1422 If I change the test to also catch `ArrowCancelled` in addition to KeyboardInterrupt, it still does not pass, because the current code code to detect that the "correct" error was raised and breaks from the loop no longer does the trick: https://github.com/apache/arrow/blob/e62ec62e40b04b0bfce76d58369845d3aa96a419/python/pyarrow/tests/test_csv.py#L1433-L1439 Because `exc_info.__context__` is now None, in contrast to what the test expects. I am not super familiar with the code that translates C++ exceptions into pyarrow/python exceptions, so not directly an idea why this PR affects that. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
