Github user mgaido91 commented on the issue:
https://github.com/apache/spark/pull/22029
> Can you put the general definition of the IN behavior (without subquery)
in the PR description?
Sure, I am changing it. Let me know if the description I am writing is
clear/complete enough. Thanks.
> nested fields are not mentioned.
They are not mentioned as in other RDBMS are not really possible. Neither
Postgres nor Oracle allow that. So I haven't changed their handling as - I
think - the expected behavior in that case is questionable. I think in general,
IN using literal and IN using subqueries should return the same result if they
are provided with the same data (as it happens in the RDBMS). So that case is
handled as before the PR since the comparison of structs has not been changed
here.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
