Shohei Okumiya

Let me close this one because we will likely go with #5077 or another approach.

Looks good to me. Note that you finally need approval from a committer

Googling some database systems, I didn't find ones with YESTERDAY or a similar UDF. But I imagine this UDF would be helpful at this point. I'm not confident though.

@MRKKmrkk Thanks for sharing the use case. I'm putting some thoughts here. 1. CI is failing. Could you please check [this](http://ci.hive.apache.org/blue/rest/organizations/jenkins/pipelines/hive-precommit/branches/PR-4985/runs/4/nodes/520/log/?start=0) and [this](http://ci.hive.apache.org/blue/rest/organizations/jenkins/pipelines/hive-precommit/branches/PR-4985/runs/4/nodes/567/log/?start=0)? I guess you need to generate the...

We may need to fix some tests

> In the PR commit build I am getting this error, but I have not changed anything in TestMiniLlapLocalCliDriver. My changes are only related to schema and I'm unable to...

@kasakrisz Thanks for checking. I double-checked the SQL:2023 Part 2. (A) `10.9 -> Syntax Rules -> 7) -> g)` mentions the specification of SUM and AVG. It says the type...

Thank you both. I believe we strongly agree with some points. - We are not motivated to keep the current behaviors of the non-CBO mode. Those of CBO are rather...

This table illustrates the summary of the current behavior tested by https://github.com/apache/hive/pull/5091/commits/468ea05ea4dc5b5ff93e7417653fe663b098a4bf. I will add my proposal in the next comment. | | CBO | Non-CBO | Note | |-|-|-|-|...

# Digest of our discussions Let me clarify the requirements of HIVE-28082. - Any SQLs have to return the same results regardless of HiveAggregateReduceFunctionsRule - $SUM0/AVG/STDDEV_(POP|SAMP)/VAR_(POP/SAMP) have to accept non...