[
https://issues.apache.org/jira/browse/CALCITE-4251?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17906848#comment-17906848
]
Stamatis Zampetakis commented on CALCITE-4251:
----------------------------------------------
Hey [~xzh_dz], it seems that the changes that were done as part of the PR#2189
in Lopt*Join classes changed unintentionally the behavior of the rules. Can you
clarify what was the original goal for adding a disjunction with
!colOrigin.isDerived() in these classes just to be 100% sure that the fix in
CALCITE-6737 is sound.
> Get the origin column, even if it is derived
> --------------------------------------------
>
> Key: CALCITE-4251
> URL: https://issues.apache.org/jira/browse/CALCITE-4251
> Project: Calcite
> Issue Type: Wish
> Reporter: xzh_dz
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.27.0
>
> Time Spent: 3.5h
> Remaining Estimate: 0h
>
> A case:
> {code:java}
> final String sql1 = ""
> + "select empno, sum(sal) as all_sal\n"
> + "from emp\n"
> + "group by empno";
> {code}
> When i try to get the `all_sal` origin column field,it will return null and
> it is derived. we always get the origin column although it is derived. We
> should overload this method and determine whether the column is derived by
> setting parameters.
> org.apache.calcite.rel.metadata.RelMetadataQuery#getColumnOrigin
> {code:java}
> // code placeholder
> public RelColumnOrigin getColumnOrigin(RelNode rel, int column) {
> final Set<RelColumnOrigin> origins = getColumnOrigins(rel, column);
> if (origins == null || origins.size() != 1) {
> return null;
> }
> final RelColumnOrigin origin = Iterables.getOnlyElement(origins);
> return origin.isDerived() ? null : origin;
> }
> {code}
>
>
>
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
