Fix Join Problem With VARCHAR Partition Columns

This change fixes the problem with joins with VARCHAR columns. Prior to
this change, when we tried to do large table joins on varchar columns, we got
an error of the form:
ERROR: cannot perform local joins that involve expressions
DETAIL: local joins can be performed between columns only.

This is because we have a check in CheckJoinBetweenColumns() which requires the
join clause to have only 'Var' nodes (i.e. columns). Postgres adds a relabel t
ype cast to cast the varchar to text; hence the type of the node is not T_Var
and the join fails.

The fix involves calling strip_implicit_coercions() to the left and right
arguments so that RELABELTYPE is stripped to VAR.

Fixes #76.

src/backend/distributed/planner/multi_physical_planner.c

@@ -3500,9 +3500,11 @@ CheckJoinBetweenColumns(OpExpr *joinClause)
 	List *argumentList = joinClause->args;
 	Node *leftArgument = (Node *) linitial(argumentList);
 	Node *rightArgument = (Node *) lsecond(argumentList);
+	Node *strippedLeftArgument = strip_implicit_coercions(leftArgument);
+	Node *strippedRightArgument = strip_implicit_coercions(rightArgument);
 
-	NodeTag leftArgumentType = nodeTag(leftArgument);
-	NodeTag rightArgumentType = nodeTag(rightArgument);
+	NodeTag leftArgumentType = nodeTag(strippedLeftArgument);
+	NodeTag rightArgumentType = nodeTag(strippedRightArgument);
 
 	if (leftArgumentType != T_Var || rightArgumentType != T_Var)
 	{

src/test/regress/expected/multi_join_pruning.out

@@ -89,13 +89,15 @@ DEBUG:  join prunable for intervals [(a,3,b),(b,4,c)] and [(c,5,d),(d,6,e)]
  explain statements for distributed queries are currently unsupported
 (1 row)
 
--- Large table joins between varchar columns do not work because of a bug we
--- have. Currently, we require joins to be only on columns. Postgres adds a
--- relabel to typecast varchars to text due to which our check fails and we
--- error out.
+-- Test that large table joins on partition varchar columns work
 EXPLAIN SELECT count(*)
 	FROM varchar_partitioned_table table1, varchar_partitioned_table table2
 	WHERE table1.varchar_column = table2.varchar_column;
-ERROR:  cannot perform local joins that involve expressions
-DETAIL:  local joins can be performed between columns only
+DEBUG:  join prunable for intervals [BA1000U2AMO4ZGX,BZZXSP27F21T6] and [AA1000U2AMO4ZGX,AZZXSP27F21T6]
+DEBUG:  join prunable for intervals [AA1000U2AMO4ZGX,AZZXSP27F21T6] and [BA1000U2AMO4ZGX,BZZXSP27F21T6]
+                              QUERY PLAN                              
+----------------------------------------------------------------------
+ explain statements for distributed queries are currently unsupported
+(1 row)
+
 SET client_min_messages TO NOTICE;

src/test/regress/sql/multi_join_pruning.sql

@@ -44,10 +44,7 @@ EXPLAIN SELECT count(*)
 	FROM composite_partitioned_table table1, composite_partitioned_table table2
 	WHERE table1.composite_column = table2.composite_column;
 
--- Large table joins between varchar columns do not work because of a bug we
--- have. Currently, we require joins to be only on columns. Postgres adds a
--- relabel to typecast varchars to text due to which our check fails and we
--- error out.
+-- Test that large table joins on partition varchar columns work
 
 EXPLAIN SELECT count(*)
 	FROM varchar_partitioned_table table1, varchar_partitioned_table table2