Github user HyukjinKwon commented on the issue: https://github.com/apache/spark/pull/20909 Yea, I know the hidden output in the console and I believe that's a known issue. In my case, I made such change before - https://github.com/apache/spark/pull/20487. Also see the discussion in https://github.com/apache/spark/pull/20465. The thing is, it needs duplicated changes to print out the warnings and that's why I have been hesitant to fix related code paths. Actually, I was thinking we should resemble what we do in `streaming.py` to skip the doctests although I haven't taken a close look to check if we can control function level yet. I know we use `# doctest: +SKIP` here and there in particular with Pandas / Arrow. I think basically we should remove this and do the same things to test them when possible. I am sure on this too (and told few committers before that I am thinking in this way). Let me cc @cloud-fan, @ueshin and @BryanCutler FYI. For the best, can you investigate and try to explicitly skip some doctests conditionally? For console output from our test script, I think we can do this separately (but please leave a comment as a todo or JIRA).
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org