[
https://issues.apache.org/jira/browse/TRAFODION-2270?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15616479#comment-15616479
]
ASF GitHub Bot commented on TRAFODION-2270:
-------------------------------------------
Github user sureshsubbiah commented on a diff in the pull request:
https://github.com/apache/incubator-trafodion/pull/795#discussion_r85603866
--- Diff: core/sql/optimizer/RelGrby.h ---
@@ -552,7 +560,9 @@ class GroupByAgg : public RelExpr
// ---------------------------------------------------------------------
ItemExpr * groupExprTree_;
ValueIdSet groupExpr_;
-
+
+ ItemExpr * extraGrpOrderby_;
+ ValueIdSet extraOrderExpr_;
--- End diff --
I am confused how an order expression can be stored in a ValueIdSet. Should
this be a ValueIdList instead?
The order (A,B) is different from (B,A) but both would be stored in the
same ValueIdSet.
This same question applies to change in OptPhysRelExpr.cpp. There we take a
ValueIdSet, convert it to a List and make a order requirement out of it. If
order is important maybe it should start as List. If order is not important
then we could add an arrangement requirement instead of a order requirement.
> support ORDER BY clause in GROUP_CONCAT function
> ------------------------------------------------
>
> Key: TRAFODION-2270
> URL: https://issues.apache.org/jira/browse/TRAFODION-2270
> Project: Apache Trafodion
> Issue Type: Sub-task
> Reporter: liu ming
> Assignee: liu ming
>
> currently, GROUP_CONCAT() in Trafodion cannot recognize order by clause, need
> to support it.
> for example:
> SELECT student_name,
> GROUP_CONCAT(DISTINCT test_score
> ORDER BY test_score asc SEPARATOR ' ')
> FROM student
> GROUP BY student_name;
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
