Merge pull request #335 from citusdata/check_count_distinct_on_single_subquery

Check count distinct on single subquery

src/backend/distributed/planner/multi_logical_optimizer.c

@@ -259,6 +259,7 @@ MultiLogicalPlanOptimize(MultiTreeRoot *multiLogicalPlan)
 		MultiTable *tableNode = (MultiTable *) lfirst(tableNodeCell);
 		if (tableNode->relationId == SUBQUERY_RELATION_ID)
 		{
+			ErrorIfContainsUnsupportedAggregate((MultiNode *) tableNode);
 			TransformSubqueryNode(tableNode);
 		}
 	}
@@ -2145,8 +2146,9 @@ ErrorIfUnsupportedAggregateDistinct(Aggref *aggregateExpression,
 	bool distinctSupported = true;
 	List *repartitionNodeList = NIL;
 	Var *distinctColumn = NULL;
-
-	AggregateType aggregateType = GetAggregateType(aggregateExpression->aggfnoid);
+	List *multiTableNodeList = NIL;
+	ListCell *multiTableNodeCell = NULL;
+	AggregateType aggregateType = AGGREGATE_INVALID_FIRST;
 
 	/* check if logical plan includes a subquery */
 	List *subqueryMultiTableList = SubqueryMultiTableList(logicalPlanNode);
@@ -2157,7 +2159,20 @@ ErrorIfUnsupportedAggregateDistinct(Aggref *aggregateExpression,
 						errdetail("distinct in the outermost query is unsupported")));
 	}
 
+	multiTableNodeList = FindNodesOfType(logicalPlanNode, T_MultiTable);
+	foreach(multiTableNodeCell, multiTableNodeList)
+	{
+		MultiTable *multiTable = (MultiTable *) lfirst(multiTableNodeCell);
+		if (multiTable->relationId == SUBQUERY_RELATION_ID)
+		{
+			ereport(ERROR, (errmsg("cannot compute count (distinct)"),
+							errdetail("Subqueries with aggregate (distinct) are "
+									  "not supported yet")));
+		}
+	}
+
 	/* if we have a count(distinct), and distinct approximation is enabled */
+	aggregateType = GetAggregateType(aggregateExpression->aggfnoid);
 	if (aggregateType == AGGREGATE_COUNT &&
 		CountDistinctErrorRate != DISABLE_DISTINCT_APPROXIMATION)
 	{

src/test/regress/expected/multi_single_relation_subquery.out

@@ -171,6 +171,18 @@ from
         l_tax) as distributed_table;
 ERROR:  cannot perform distributed planning on this query
 DETAIL:  Subqueries without aggregates are not supported yet
+-- Check that we don't support subqueries with count(distinct).
+select
+    different_shipment_days
+from
+    (select
+        count(distinct l_shipdate) as different_shipment_days
+    from
+        lineitem
+    group by
+        l_partkey) as distributed_table;
+ERROR:  cannot compute count (distinct)
+DETAIL:  Subqueries with aggregate (distinct) are not supported yet
 -- Check that if subquery is pulled, we don't error and run query properly.
 SELECT max(l_suppkey) FROM
 (

src/test/regress/sql/multi_single_relation_subquery.sql

@@ -125,6 +125,18 @@ from
     group by
         l_tax) as distributed_table;
 
+-- Check that we don't support subqueries with count(distinct).
+
+select
+    different_shipment_days
+from
+    (select
+        count(distinct l_shipdate) as different_shipment_days
+    from
+        lineitem
+    group by
+        l_partkey) as distributed_table;
+
 -- Check that if subquery is pulled, we don't error and run query properly.
 
 SELECT max(l_suppkey) FROM