[
https://issues.apache.org/jira/browse/CALCITE-734?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16845462#comment-16845462
]
Feng Zhu commented on CALCITE-734:
----------------------------------
Hi, [~julianhyde]. We use Calcite to pre-process (i.e., parse, validate,
convert and optimize) SQL query to RelNode, and then convert RelNode back to
SQL queries to be executed in different engines. However, we found
_*RelToSqlConverter*_ generates wrong queries for "Aggregate.groupSets contains
more than 1 element". Consequently, *CUBE/ROLLUP/GROUPING SETS* clauses fail to
be transformed.
I track the problem and find several JIRAs you created. It seems to be some
legacy issues, e.g., JdbcAggregateRule, group_id() and so on. I try to fix
these issues. At the beginning, I plan to convert an Aggregate with
multi-groupSets into Union ALL in _*RelToSqlConvertor*_.
Looking forward to your suggestions.
> Push GROUPING SETS to underlying SQL via JDBC adapter
> -----------------------------------------------------
>
> Key: CALCITE-734
> URL: https://issues.apache.org/jira/browse/CALCITE-734
> Project: Calcite
> Issue Type: Bug
> Components: jdbc-adapter
> Reporter: Julian Hyde
> Priority: Major
>
> Push GROUPING SETS to underlying SQL via JDBC adapter. Currently
> JdbcAggregateRule refuses if Aggregate.groupSets contains more than 1 element.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
