[
https://issues.apache.org/jira/browse/CALCITE-7493?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18076865#comment-18076865
]
Julian Hyde commented on CALCITE-7493:
--------------------------------------
It would be valid to optimize “count(distinct sal)” to “case when sal is null
then 0 else 1 end”, and I believe we already do that. But optimizing
“count(sal)” to “1” in that query doesn’t seem valid.
> Support constant-result aggregates (e.g., COUNT, STDDEV) over GROUP BY keys
> ---------------------------------------------------------------------------
>
> Key: CALCITE-7493
> URL: https://issues.apache.org/jira/browse/CALCITE-7493
> Project: Calcite
> Issue Type: Improvement
> Components: core
> Affects Versions: 1.41.0
> Reporter: Yu Xu
> Assignee: Yu Xu
> Priority: Major
>
> Sql like：
> {code:java}
> select sal, count(sal) as cnt, stddev(sal) as sd,
> stddev_pop(sal) as sdp, var_pop(sal) as vp, var_samp(sal) as vs
> from emp group by sal, deptno {code}
>
> It should be able to be optimized to:
> ||agg funcions||Better simplified results||
> |{{COUNT(sal)}}|{{1}}|
> |{{STDDEV(sal)}}|{{0}}|
> |{{STDDEV_POP(sal)}}|{{0}}|
> and current plan:
> {code:java}
> LogicalProject(SAL=[$0], CNT=[$2], SD=[$3], SDP=[$4], VP=[$5], VS=[$6])
> LogicalAggregate(group=[{0, 1}], CNT=[COUNT()], SD=[STDDEV($0)],
> SDP=[STDDEV_POP($0)], VP=[VAR_POP($0)], VS=[VAR_SAMP($0)])
> LogicalProject(SAL=[$5], DEPTNO=[$7])
> LogicalTableScan(table=[[CATALOG, SALES, EMP]]) {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
