Adjust the existing regression tests

src/backend/distributed/planner/query_colocation_checker.c

@@ -21,6 +21,13 @@
 
 #include "postgres.h"
 
+#include "distributed/pg_version_constants.h"
+
+#if PG_VERSION_NUM >= PG_VERSION_12
+#include "access/relation.h"
+#else
+#include "access/heapam.h"
+#endif
 #include "distributed/multi_logical_planner.h"
 #include "distributed/query_colocation_checker.h"
 #include "distributed/pg_dist_partition.h"
@@ -269,7 +276,6 @@ WrapRteRelationIntoSubquery(RangeTblEntry *rteRelation)
 	newRangeTableRef->rtindex = 1;
 	subquery->jointree = makeFromExpr(list_make1(newRangeTableRef), NULL);
 
-
 	Relation relation = relation_open(rteRelation->relid, AccessShareLock);
 	int numberOfAttributes = RelationGetNumberOfAttributes(relation);
 

src/backend/distributed/planner/relation_restriction_equivalence.c

@@ -30,9 +30,9 @@
 #else
 #include "optimizer/cost.h"
 #include "nodes/relation.h"
+#include "optimizer/var.h"
 #endif
 #include "optimizer/paths.h"
-#include "optimizer/var.h"
 #include "parser/parsetree.h"
 #include "optimizer/pathnode.h"
 
@@ -1879,14 +1879,9 @@ GetRestrictInfoListForRelation(RangeTblEntry *rangeTblEntry,
 	{
 		RestrictInfo *restrictInfo = (RestrictInfo *) lfirst(restrictCell);
 		Expr *restrictionClause = restrictInfo->clause;
-		List *varClauses = NIL;
-		ListCell *varClauseCell = NULL;
-		Relids varnos = NULL;
-
-		Expr *copyOfRestrictClause = NULL;
 
 		/* we cannot process Params beacuse they are not known at this point */
-		if (FindNodeCheck((Node *) restrictionClause, IsParam))
+		if (FindNodeMatchingCheckFunction((Node *) restrictionClause, IsParam))
 		{
 			continue;
 		}
@@ -1895,7 +1890,7 @@ GetRestrictInfoListForRelation(RangeTblEntry *rangeTblEntry,
 		 * If the restriction involves multiple tables, we cannot add it to
 		 * input relation's expression list.
 		 */
-		varnos = pull_varnos((Node *) restrictionClause);
+		Relids varnos = pull_varnos((Node *) restrictionClause);
 		if (bms_num_members(varnos) != 1)
 		{
 			continue;
@@ -1906,14 +1901,15 @@ GetRestrictInfoListForRelation(RangeTblEntry *rangeTblEntry,
 		 * which consists of only one relation in its jointree. Thus,
 		 * simply set the varnos accordingly.
 		 */
-		copyOfRestrictClause = (Expr *) copyObject((Node *) restrictionClause);
-		varClauses = pull_var_clause_default((Node *) copyOfRestrictClause);
+		Expr *copyOfRestrictClause = (Expr *) copyObject((Node *) restrictionClause);
+		List *varClauses = pull_var_clause_default((Node *) copyOfRestrictClause);
+		ListCell *varClauseCell = NULL;
 		foreach(varClauseCell, varClauses)
 		{
 			Var *column = (Var *) lfirst(varClauseCell);
 
 			column->varno = rteIndex;
-			column->varnoold = rteIndex;
+			column->varnosyn = rteIndex;
 		}
 
 		restrictExprList = lappend(restrictExprList, copyOfRestrictClause);

src/test/regress/expected/coordinator_shouldhaveshards.out

@@ -417,12 +417,22 @@ HINT:  To remove the local data, run: SELECT truncate_local_data_after_distribut
 
 INSERT INTO dist_table VALUES(1);
 NOTICE:  executing the command locally: INSERT INTO coordinator_shouldhaveshards.dist_table_1503017 (a) VALUES (1)
-SELECT * FROM local JOIN dist_table ON (a = x);
-ERROR:  direct joins between distributed and local tables are not supported
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
-SELECT * FROM local JOIN dist_table ON (a = x) WHERE a = 1;;
-ERROR:  direct joins between distributed and local tables are not supported
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
+SELECT * FROM local JOIN dist_table ON (a = x) ORDER BY 1,2,3;
+ x | y | a
+---------------------------------------------------------------------
+ 1 | 2 | 1
+ 1 | 2 | 1
+ 3 | 2 | 3
+(3 rows)
+
+SELECT * FROM local JOIN dist_table ON (a = x) WHERE a = 1 ORDER BY 1,2,3;
+NOTICE:  executing the command locally: SELECT local.x, local.y, dist_table.a FROM ((SELECT intermediate_result.x, intermediate_result.y FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(x integer, y integer)) local JOIN coordinator_shouldhaveshards.dist_table_1503017 dist_table ON ((dist_table.a OPERATOR(pg_catalog.=) local.x))) WHERE (dist_table.a OPERATOR(pg_catalog.=) 1) ORDER BY local.x, local.y, dist_table.a
+ x | y | a
+---------------------------------------------------------------------
+ 1 | 2 | 1
+ 1 | 2 | 1
+(2 rows)
+
 -- intermediate results are allowed
 WITH cte_1 AS (SELECT * FROM dist_table ORDER BY 1 LIMIT 1)
 SELECT * FROM ref JOIN local ON (a = x) JOIN cte_1 ON (local.x = cte_1.a);

src/test/regress/expected/dml_recursive.out

@@ -340,8 +340,10 @@ FROM cte_1
 WHERE distributed_table.tenant_id < cte_1.tenant_id;
 DEBUG:  generating subplan XXX_1 for CTE cte_1: WITH cte_2 AS (SELECT second_distributed_table.tenant_id AS cte2_id FROM recursive_dml_queries.second_distributed_table WHERE (second_distributed_table.dept OPERATOR(pg_catalog.>=) 2)) UPDATE recursive_dml_queries.distributed_table SET dept = 10 RETURNING tenant_id, dept, info
 DEBUG:  Plan XXX query after replacing subqueries and CTEs: UPDATE recursive_dml_queries.distributed_table SET dept = 5 FROM (SELECT intermediate_result.tenant_id, intermediate_result.dept, intermediate_result.info FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(tenant_id text, dept integer, info jsonb)) cte_1 WHERE (distributed_table.tenant_id OPERATOR(pg_catalog.<) cte_1.tenant_id)
--- we don't support updating local table with a join with
--- distributed tables
+-- we support updating local table with a join with
+-- distributed tables, though as the local table
+-- is target here, distributed table is recursively
+-- planned
 UPDATE
 	local_table
 SET
@@ -350,8 +352,9 @@ FROM
 	distributed_table
 WHERE
 	distributed_table.tenant_id = local_table.id;
-ERROR:  cannot plan modifications with local tables involving citus tables
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
+DEBUG:  Wrapping local relation "distributed_table" to a subquery: SELECT tenant_id, dept, info FROM recursive_dml_queries.distributed_table WHERE true OFFSET 0
+DEBUG:  generating subplan XXX_1 for subquery SELECT tenant_id, dept, info FROM recursive_dml_queries.distributed_table WHERE true OFFSET 0
+DEBUG:  Plan XXX query after replacing subqueries and CTEs: UPDATE recursive_dml_queries.local_table SET id = 'citus_test'::text FROM (SELECT intermediate_result.tenant_id, intermediate_result.dept, intermediate_result.info FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(tenant_id text, dept integer, info jsonb)) distributed_table WHERE (distributed_table.tenant_id OPERATOR(pg_catalog.=) local_table.id)
 RESET client_min_messages;
 DROP SCHEMA recursive_dml_queries CASCADE;
 NOTICE:  drop cascades to 5 other objects

src/test/regress/expected/multi_modifications.out

@@ -270,8 +270,6 @@ CREATE TABLE bidders ( name text, id bigint );
 DELETE FROM limit_orders USING bidders WHERE limit_orders.id = 246 AND
 											 limit_orders.bidder_id = bidders.id AND
 											 bidders.name = 'Bernie Madoff';
-ERROR:  cannot plan modifications with local tables involving citus tables
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
 -- commands containing a CTE are supported
 WITH new_orders AS (INSERT INTO limit_orders VALUES (411, 'FLO', 12, '2017-07-02 16:32:15', 'buy', 66))
 DELETE FROM limit_orders WHERE id < 0;
@@ -429,13 +427,11 @@ ERROR:  modifying the partition value of rows is not allowed
 UPDATE limit_orders SET id = 246 WHERE id = 246;
 UPDATE limit_orders SET id = 246 WHERE id = 246 AND symbol = 'GM';
 UPDATE limit_orders SET id = limit_orders.id WHERE id = 246;
--- UPDATEs with a FROM clause are unsupported
+-- UPDATEs with a FROM clause are supported even with local tables
 UPDATE limit_orders SET limit_price = 0.00 FROM bidders
 					WHERE limit_orders.id = 246 AND
 						  limit_orders.bidder_id = bidders.id AND
 						  bidders.name = 'Bernie Madoff';
-ERROR:  cannot plan modifications with local tables involving citus tables
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
 -- should succeed with a CTE
 WITH deleted_orders AS (INSERT INTO limit_orders VALUES (399, 'PDR', 14, '2017-07-02 16:32:15', 'sell', 43))
 UPDATE limit_orders SET symbol = 'GM';
@@ -1306,7 +1302,8 @@ DELETE FROM summary_table WHERE id < (
 );
 CREATE TABLE multi_modifications.local (a int default 1, b int);
 INSERT INTO multi_modifications.local VALUES (default, (SELECT min(id) FROM summary_table));
-ERROR:  cannot plan modifications of local tables involving distributed tables
+ERROR:  subqueries are not supported within INSERT queries
+HINT:  Try rewriting your queries with 'INSERT INTO ... SELECT' syntax.
 DROP TABLE raw_table;
 DROP TABLE summary_table;
 DROP TABLE reference_raw_table;

src/test/regress/expected/multi_mx_modifications.out

@@ -158,7 +158,6 @@ CREATE TABLE bidders ( name text, id bigint );
 DELETE FROM limit_orders_mx USING bidders WHERE limit_orders_mx.id = 246 AND
 											 limit_orders_mx.bidder_id = bidders.id AND
 											 bidders.name = 'Bernie Madoff';
-ERROR:  cannot plan modifications with local tables involving citus tables
 -- commands containing a CTE are supported
 WITH new_orders AS (INSERT INTO limit_orders_mx VALUES (411, 'FLO', 12, '2017-07-02 16:32:15', 'buy', 66))
 DELETE FROM limit_orders_mx WHERE id < 0;
@@ -225,7 +224,6 @@ UPDATE limit_orders_mx SET limit_price = 0.00 FROM bidders
 					WHERE limit_orders_mx.id = 246 AND
 						  limit_orders_mx.bidder_id = bidders.id AND
 						  bidders.name = 'Bernie Madoff';
-ERROR:  cannot plan modifications with local tables involving citus tables
 -- commands containing a CTE are supported
 WITH deleted_orders AS (INSERT INTO limit_orders_mx VALUES (399, 'PDR', 14, '2017-07-02 16:32:15', 'sell', 43))
 UPDATE limit_orders_mx SET symbol = 'GM';

src/test/regress/expected/multi_shard_modify.out

@@ -72,8 +72,11 @@ CREATE TABLE temp_nations(name text, key integer);
 SELECT master_modify_multiple_shards('DELETE FROM multi_shard_modify_test USING temp_nations WHERE multi_shard_modify_test.t_value = temp_nations.key AND temp_nations.name = ''foobar'' ');
 WARNING:  master_modify_multiple_shards is deprecated and will be removed in a future release.
 HINT:  Run the command directly
-ERROR:  cannot plan modifications with local tables involving citus tables
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
+ master_modify_multiple_shards
+---------------------------------------------------------------------
+                             0
+(1 row)
+
 -- commands with a USING clause are unsupported
 SELECT create_distributed_table('temp_nations', 'name', 'hash');
  create_distributed_table

src/test/regress/expected/multi_shard_update_delete.out

@@ -730,19 +730,15 @@ SET    value_2 = subquery.random FROM (SELECT user_id, random()
 WHERE  users_test_table.user_id = subquery.user_id;
 -- Make following tests consistent
 UPDATE users_test_table SET value_2 = 0;
--- Local tables are not supported
+-- Joins with tables not supported
 UPDATE users_test_table
 SET    value_2 = 5
 FROM   events_test_table_local
 WHERE  users_test_table.user_id = events_test_table_local.user_id;
-ERROR:  cannot plan modifications with local tables involving citus tables
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
 UPDATE events_test_table_local
 SET    value_2 = 5
 FROM   users_test_table
 WHERE  events_test_table_local.user_id = users_test_table.user_id;
-ERROR:  cannot plan modifications with local tables involving citus tables
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
 -- Local tables in a subquery are supported through recursive planning
 UPDATE users_test_table
 SET    value_2 = 5

src/test/regress/expected/multi_simple_queries.out

@@ -280,15 +280,20 @@ ORDER BY articles.id;
 -- subqueries are not supported in SELECT clause
 SELECT a.title AS name, (SELECT a2.id FROM articles_single_shard a2 WHERE a.id = a2.id  LIMIT 1)
 						 AS special_price FROM articles a;
-ERROR:  complex joins are only supported when all distributed tables are co-located and joined on their distribution columns
--- joins are not supported between local and distributed tables
+ERROR:  could not run distributed query with subquery outside the FROM, WHERE and HAVING clauses
+HINT:  Consider using an equality filter on the distributed table's partition column.
+-- joins are supported between local and distributed tables
 SELECT title, authors.name FROM authors, articles WHERE authors.id = articles.author_id;
-ERROR:  direct joins between distributed and local tables are not supported
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
--- inner joins are not supported (I think)
+ title | name
+---------------------------------------------------------------------
+(0 rows)
+
+-- inner joins are supported
 SELECT * FROM  (articles INNER JOIN authors ON articles.id = authors.id);
-ERROR:  direct joins between distributed and local tables are not supported
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
+ id | author_id | title | word_count | name | id
+---------------------------------------------------------------------
+(0 rows)
+
 -- test use of EXECUTE statements within plpgsql
 DO $sharded_execute$
 	BEGIN

src/test/regress/expected/multi_simple_queries_0.out

@@ -226,12 +226,12 @@ SELECT a.title AS name, (SELECT a2.id FROM articles_single_shard a2 WHERE a.id =
 						 AS special_price FROM articles a;
 ERROR:  could not run distributed query with subquery outside the FROM, WHERE and HAVING clauses
 HINT:  Consider using an equality filter on the distributed table's partition column.
--- joins are not supported between local and distributed tables
+-- joins are supported between local and distributed tables
 SELECT title, authors.name FROM authors, articles WHERE authors.id = articles.author_id;
-ERROR:  relation authors is not distributed
--- inner joins are not supported (I think)
+ERROR:  Complex subqueries and CTEs are not supported when task_executor_type is set to 'task-tracker'
+-- inner joins are supported
 SELECT * FROM  (articles INNER JOIN authors ON articles.id = authors.id);
-ERROR:  relation authors is not distributed
+ERROR:  Complex subqueries and CTEs are not supported when task_executor_type is set to 'task-tracker'
 -- test use of EXECUTE statements within plpgsql
 DO $sharded_execute$
 	BEGIN

src/test/regress/expected/with_executors.out

@@ -302,13 +302,16 @@ SELECT min(user_id) FROM cte JOIN local_table ON (user_id = id);
    1
 (1 row)
 
--- not if there are no distributed tables
+-- even if there are no distributed tables
 WITH cte AS (
 	SELECT user_id FROM users_table
 )
 SELECT min(user_id) FROM cte JOIN local_table ON (user_id = id) JOIN events_table USING (user_id);
-ERROR:  direct joins between distributed and local tables are not supported
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
+ min
+---------------------------------------------------------------------
+   1
+(1 row)
+
 -- unless the distributed table is part of a recursively planned subquery
 WITH cte AS (
 	SELECT user_id FROM users_table
@@ -319,15 +322,18 @@ SELECT min(user_id) FROM cte JOIN local_table ON (user_id = id) JOIN (SELECT * F
    1
 (1 row)
 
--- joins between local and reference tables not allowed
--- since the coordinator is not in the metadata at this stage
+-- joins between local and reference tables are allowed
+-- even when the coordinator is not in the metadata at this stage
 WITH cte AS (
 	SELECT user_id FROM users_table
 )
 SELECT count(*) FROM local_table JOIN ref_table USING (id)
 WHERE id IN (SELECT * FROM cte);
-ERROR:  direct joins between distributed and local tables are not supported
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
+ count
+---------------------------------------------------------------------
+     6
+(1 row)
+
 -- CTEs should be able to terminate a router query
 WITH cte AS (
 	WITH cte_1 AS (

src/test/regress/expected/with_modifying.out

@@ -668,19 +668,20 @@ SELECT * FROM raw_data ORDER BY val;
 ---------------------------------------------------------------------
 (0 rows)
 
--- Test that local tables are barred
+-- Test that local tables are can be updated
+-- selecting from distributed tables
 UPDATE local_table lt SET val = mt.val
 FROM modify_table mt WHERE mt.id = lt.id;
-ERROR:  cannot plan modifications with local tables involving citus tables
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
 -- Including inside CTEs
 WITH cte AS (
 	UPDATE local_table lt SET val = mt.val
 	FROM modify_table mt WHERE mt.id = lt.id
 	RETURNING lt.id, lt.val
 ) SELECT * FROM cte JOIN modify_table mt ON mt.id = cte.id ORDER BY 1,2;
-ERROR:  cannot plan modifications with local tables involving citus tables
-HINT:  Use CTE's or subqueries to select from local tables and use them in joins
+ id | val | id | val
+---------------------------------------------------------------------
+(0 rows)
+
 -- Make sure checks for volatile functions apply to CTEs too
 WITH cte AS (UPDATE modify_table SET val = random() WHERE id = 3 RETURNING *)
 SELECT * FROM cte JOIN modify_table mt ON mt.id = 3 AND mt.id = cte.id ORDER BY 1,2;

src/test/regress/sql/coordinator_shouldhaveshards.sql

@@ -169,8 +169,8 @@ CREATE TABLE dist_table(a int);
 SELECT create_distributed_table('dist_table', 'a');
 INSERT INTO dist_table VALUES(1);
 
-SELECT * FROM local JOIN dist_table ON (a = x);
-SELECT * FROM local JOIN dist_table ON (a = x) WHERE a = 1;;
+SELECT * FROM local JOIN dist_table ON (a = x) ORDER BY 1,2,3;
+SELECT * FROM local JOIN dist_table ON (a = x) WHERE a = 1 ORDER BY 1,2,3;
 
 -- intermediate results are allowed
 WITH cte_1 AS (SELECT * FROM dist_table ORDER BY 1 LIMIT 1)

src/test/regress/sql/dml_recursive.sql

@@ -281,8 +281,10 @@ SET dept = 5
 FROM cte_1
 WHERE distributed_table.tenant_id < cte_1.tenant_id;
 
--- we don't support updating local table with a join with
--- distributed tables
+-- we support updating local table with a join with
+-- distributed tables, though as the local table
+-- is target here, distributed table is recursively
+-- planned
 UPDATE
 	local_table
 SET

src/test/regress/sql/multi_modifications.sql

@@ -327,7 +327,7 @@ UPDATE limit_orders SET id = 246 WHERE id = 246;
 UPDATE limit_orders SET id = 246 WHERE id = 246 AND symbol = 'GM';
 UPDATE limit_orders SET id = limit_orders.id WHERE id = 246;
 
--- UPDATEs with a FROM clause are unsupported
+-- UPDATEs with a FROM clause are supported even with local tables
 UPDATE limit_orders SET limit_price = 0.00 FROM bidders
 					WHERE limit_orders.id = 246 AND
 						  limit_orders.bidder_id = bidders.id AND

src/test/regress/sql/multi_shard_update_delete.sql

@@ -609,7 +609,7 @@ WHERE  users_test_table.user_id = subquery.user_id;
 -- Make following tests consistent
 UPDATE users_test_table SET value_2 = 0;
 
--- Local tables are not supported
+-- Joins with tables not supported
 UPDATE users_test_table
 SET    value_2 = 5
 FROM   events_test_table_local

src/test/regress/sql/multi_simple_queries.sql

@@ -146,10 +146,10 @@ ORDER BY articles.id;
 SELECT a.title AS name, (SELECT a2.id FROM articles_single_shard a2 WHERE a.id = a2.id  LIMIT 1)
 						 AS special_price FROM articles a;
 
--- joins are not supported between local and distributed tables
+-- joins are supported between local and distributed tables
 SELECT title, authors.name FROM authors, articles WHERE authors.id = articles.author_id;
 
--- inner joins are not supported (I think)
+-- inner joins are supported
 SELECT * FROM  (articles INNER JOIN authors ON articles.id = authors.id);
 
 -- test use of EXECUTE statements within plpgsql

src/test/regress/sql/with_executors.sql

@@ -231,7 +231,7 @@ WITH cte AS (
 )
 SELECT min(user_id) FROM cte JOIN local_table ON (user_id = id);
 
--- not if there are no distributed tables
+-- even if there are no distributed tables
 WITH cte AS (
 	SELECT user_id FROM users_table
 )
@@ -243,8 +243,8 @@ WITH cte AS (
 )
 SELECT min(user_id) FROM cte JOIN local_table ON (user_id = id) JOIN (SELECT * FROM events_table OFFSET 0) e USING (user_id);
 
--- joins between local and reference tables not allowed
--- since the coordinator is not in the metadata at this stage
+-- joins between local and reference tables are allowed
+-- even when the coordinator is not in the metadata at this stage
 WITH cte AS (
 	SELECT user_id FROM users_table
 )

src/test/regress/sql/with_modifying.sql

@@ -412,7 +412,8 @@ raw_data AS (
 )
 SELECT * FROM raw_data ORDER BY val;
 
--- Test that local tables are barred
+-- Test that local tables are can be updated
+-- selecting from distributed tables
 UPDATE local_table lt SET val = mt.val
 FROM modify_table mt WHERE mt.id = lt.id;