Github user nsyca commented on the issue:
https://github.com/apache/spark/pull/16337
@hvanhovell: On the correctness, how about running the same set of test
cases against a second SQL system and comparing the results? If the two systems
yield the same result, we can assume Spark has the correct result.
Besides the correctness, what aspect of review would you be thinking? Areas
of test coverage? Two more sets of test cases for EXISTS and scalar subquery
will be submitted in the coming week. I want to make sure we will follow the
same process.
Note that there are negative test cases (test cases that currently raise
Analysis exceptions) we have not included in this PR. We plan to put those in a
separate PR with some minor code change to suppress the output of the full
LogicalPlan along with the error messages. This is because the infrastructure
of SQLQueryTestSuite might not be designed to compare two equivalent
LogicalPlans of different ExprIds.
We want to have sufficient test coverage for all the three forms of
subquery. This way when we will submit the code to defer the pull up, we can
run against these sets of test cases to mitigate any undesired effect from the
code change.
I don't understand your last point (are the tests part of a test kit?).
Could you please elaborate?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
