Github user mgaido91 commented on the issue:
https://github.com/apache/spark/pull/21054
I added the sql tests as you suggested @maropu. Doing this, I found that we
have some difference in the way we return `NaN` and `null` wrt Hive. I haven't
changed our behavior though, since this would introduce a behavior change for
methods which already exist (corr).
I think we have to discuss this and eventually fix in a separate PR if you
agree, since this is not really related to the introduction of the new
functions.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
