Add dependencies for stat schemas

src/backend/distributed/commands/statistics.c

@@ -144,10 +144,6 @@ GetExplicitStatisticsCommandList(Oid relationId)
 	Oid statisticsId = InvalidOid;
 	foreach_oid(statisticsId, statisticsIdList)
 	{
-		ObjectAddress address = { 0 };
-		ObjectAddressSet(address, StatisticExtRelationId, statisticsId);
-		EnsureDependenciesExistOnAllNodes(&address);
-
 		char *createStatisticsCommand = pg_get_statisticsobj_worker(statisticsId, false);
 
 		createStatisticsCommandList = lappend(
@@ -162,6 +158,51 @@ GetExplicitStatisticsCommandList(Oid relationId)
 }
 
 
+/*
+ * GetExplicitStatisticsSchemaIdList returns the list of schema ids of statistics'
+ * which are created on relation with given relation id.
+ */
+List *
+GetExplicitStatisticsSchemaIdList(Oid relationId)
+{
+	List *schemaIdList = NIL;
+
+	Relation pgStatistics = table_open(StatisticExtRelationId, AccessShareLock);
+
+	int scanKeyCount = 1;
+	ScanKeyData scanKey[1];
+
+	ScanKeyInit(&scanKey[0], Anum_pg_statistic_ext_stxrelid,
+				BTEqualStrategyNumber, F_OIDEQ, relationId);
+
+	bool useIndex = true;
+	SysScanDesc scanDescriptor = systable_beginscan(pgStatistics,
+													StatisticExtRelidIndexId,
+													useIndex, NULL, scanKeyCount,
+													scanKey);
+
+	HeapTuple heapTuple = systable_getnext(scanDescriptor);
+	while (HeapTupleIsValid(heapTuple))
+	{
+		FormData_pg_statistic_ext *statisticsForm =
+			(FormData_pg_statistic_ext *) GETSTRUCT(heapTuple);
+
+		Oid schemaId = statisticsForm->stxnamespace;
+		if (!list_member_oid(schemaIdList, schemaId))
+		{
+			schemaIdList = lappend_oid(schemaIdList, schemaId);
+		}
+
+		heapTuple = systable_getnext(scanDescriptor);
+	}
+
+	systable_endscan(scanDescriptor);
+	table_close(pgStatistics, NoLock);
+
+	return schemaIdList;
+}
+
+
 /*
  * GetExplicitStatisticsIdList returns a list of OIDs corresponding to the statistics
  * that are explicitly created on the relation with relationId. That means,

src/backend/distributed/commands/type.c

@@ -258,7 +258,7 @@ PreprocessCreateEnumStmt(Node *node, const char *queryString)
 	QualifyTreeNode(node);
 
 	/* reconstruct creation statement in a portable fashion */
-	const char *createEnumStmtSql = DeparseTreeNode((Node *) node);
+	const char *createEnumStmtSql = DeparseCreateEnumStmt(node);
 	createEnumStmtSql = WrapCreateOrReplace(createEnumStmtSql);
 
 	/*

src/backend/distributed/metadata/dependency.c

@@ -21,6 +21,7 @@
 #include "catalog/indexing.h"
 #include "catalog/pg_class.h"
 #include "catalog/pg_depend.h"
+#include "catalog/pg_namespace.h"
 #include "catalog/pg_proc_d.h"
 #include "catalog/pg_rewrite.h"
 #include "catalog/pg_rewrite_d.h"
@@ -119,6 +120,7 @@ typedef struct ViewDependencyNode
 
 
 static List * GetRelationTriggerFunctionDepencyList(Oid relationId);
+static List * GetRelationStatsSchemaDependencyList(Oid relationId);
 static DependencyDefinition * CreateObjectAddressDependencyDef(Oid classId, Oid objectId);
 static ObjectAddress DependencyDefinitionObjectAddress(DependencyDefinition *definition);
 
@@ -926,6 +928,17 @@ ExpandCitusSupportedTypes(ObjectAddressCollector *collector, ObjectAddress targe
 			List *triggerFunctionDepencyList =
 				GetRelationTriggerFunctionDepencyList(relationId);
 			result = list_concat(result, triggerFunctionDepencyList);
+
+			/*
+			 * Statistics' both depend to the relations and to the schemas they belong
+			 * to. Also, pg_depend records dependencies from statistics to their schemas
+			 * but not from relations to their statistics' schemas. Given above two,
+			 * we directly expand dependencies for the relations to schemas of
+			 * statistics.
+			 */
+			List *statisticsSchemaDependencyList =
+				GetRelationStatsSchemaDependencyList(relationId);
+			result = list_concat(result, statisticsSchemaDependencyList);
 		}
 
 		default:
@@ -938,6 +951,28 @@ ExpandCitusSupportedTypes(ObjectAddressCollector *collector, ObjectAddress targe
 }
 
 
+/*
+ * GetRelationStatsSchemaDependencyList returns a list of DependencyDefinition
+ * objects for the schemas that statistics' of the relation with relationId depends.
+ */
+static List *
+GetRelationStatsSchemaDependencyList(Oid relationId)
+{
+	List *dependencyList = NIL;
+
+	List *schemaIds = GetExplicitStatisticsSchemaIdList(relationId);
+	Oid schemaId = InvalidOid;
+	foreach_oid(schemaId, schemaIds)
+	{
+		DependencyDefinition *dependency =
+			CreateObjectAddressDependencyDef(NamespaceRelationId, schemaId);
+		dependencyList = lappend(dependencyList, dependency);
+	}
+
+	return dependencyList;
+}
+
+
 /*
  * GetRelationTriggerFunctionDepencyList returns a list of DependencyDefinition
  * objects for the functions that triggers of the relation with relationId depends.

src/include/distributed/commands.h

@@ -244,6 +244,7 @@ extern List * PreprocessCreateStatisticsStmt(Node *node, const char *queryString
 extern List * PostprocessCreateStatisticsStmt(Node *node, const char *queryString);
 extern ObjectAddress CreateStatisticsStmtObjectAddress(Node *node, bool missingOk);
 extern List * GetExplicitStatisticsCommandList(Oid relationId);
+extern List * GetExplicitStatisticsSchemaIdList(Oid relationId);
 
 /* subscription.c - forward declarations */
 extern Node * ProcessCreateSubscriptionStmt(CreateSubscriptionStmt *createSubStmt);

src/test/regress/expected/propagate_statistics.out

@@ -1,5 +1,5 @@
-CREATE SCHEMA "statistics'test";
+CREATE SCHEMA "statistics'Test";
-SET search_path TO "statistics'test";
+SET search_path TO "statistics'Test";
 SET citus.next_shard_id TO 980000;
 SET client_min_messages TO WARNING;
 SET citus.shard_count TO 32;
@@ -17,12 +17,12 @@ SELECT create_distributed_table('test_stats', 'a');
 
 CREATE STATISTICS s1 (dependencies) ON a, b FROM test_stats;
 -- test for distributing an already existing statistics
-CREATE TABLE test_stats2 (
+CREATE TABLE "test'stats2" (
     a   int,
     b   int
 );
-CREATE STATISTICS s2 (dependencies) ON a, b FROM test_stats2;
+CREATE STATISTICS s2 (dependencies) ON a, b FROM "test'stats2";
-SELECT create_distributed_table('test_stats2', 'a');
+SELECT create_distributed_table('test''stats2', 'a');
  create_distributed_table
 ---------------------------------------------------------------------
 
@@ -30,18 +30,54 @@ SELECT create_distributed_table('test_stats2', 'a');
 
 -- test when stats is on a different schema
 CREATE SCHEMA sc1;
-CREATE TABLE tbl (a int, b int);
+CREATE TABLE tbl (a int, "B" text);
 SELECT create_distributed_table ('tbl', 'a');
  create_distributed_table
 ---------------------------------------------------------------------
 
 (1 row)
 
-CREATE STATISTICS sc1.st1 ON a, b FROM tbl;
+CREATE STATISTICS sc1.st1 ON a, "B" FROM tbl;
-\c - - - :worker_1_port
+-- test distributing table with already created stats on a new schema
-SELECT stxname FROM pg_statistic_ext ORDER BY stxname ASC;
+CREATE TABLE test_stats3 (
-  stxname
+    a int,
+    b int
+);
+CREATE SCHEMA sc2;
+CREATE STATISTICS sc2."neW'Stat" ON a,b FROM test_stats3;
+SELECT create_distributed_table ('test_stats3','a');
+ create_distributed_table
 ---------------------------------------------------------------------
+
+(1 row)
+
+\c - - - :worker_1_port
+SELECT stxname
+FROM pg_statistic_ext
+WHERE stxnamespace IN (
+	SELECT oid
+	FROM pg_namespace
+	WHERE nspname IN ('public', 'statistics''Test', 'sc1', 'sc2')
+)
+ORDER BY stxname ASC;
+     stxname
+---------------------------------------------------------------------
+ neW'Stat_980096
+ neW'Stat_980098
+ neW'Stat_980100
+ neW'Stat_980102
+ neW'Stat_980104
+ neW'Stat_980106
+ neW'Stat_980108
+ neW'Stat_980110
+ neW'Stat_980112
+ neW'Stat_980114
+ neW'Stat_980116
+ neW'Stat_980118
+ neW'Stat_980120
+ neW'Stat_980122
+ neW'Stat_980124
+ neW'Stat_980126
  s1_980000
  s1_980002
  s1_980004
@@ -90,15 +126,22 @@ SELECT stxname FROM pg_statistic_ext ORDER BY stxname ASC;
  st1_980090
  st1_980092
  st1_980094
-(48 rows)
+(64 rows)
 
-SELECT count(DISTINCT stxnamespace) FROM pg_statistic_ext;
+SELECT count(DISTINCT stxnamespace)
+FROM pg_statistic_ext
+WHERE stxnamespace IN (
+	SELECT oid
+	FROM pg_namespace
+	WHERE nspname IN ('public', 'statistics''Test', 'sc1', 'sc2')
+);
  count
 ---------------------------------------------------------------------
-     2
+     3
 (1 row)
 
 \c - - - :master_port
 SET client_min_messages TO WARNING;
-DROP SCHEMA "statistics'test" CASCADE;
+DROP SCHEMA "statistics'Test" CASCADE;
 DROP SCHEMA sc1 CASCADE;
+DROP SCHEMA sc2 CASCADE;

src/test/regress/sql/propagate_statistics.sql

@@ -1,6 +1,6 @@
-CREATE SCHEMA "statistics'test";
+CREATE SCHEMA "statistics'Test";
 
-SET search_path TO "statistics'test";
+SET search_path TO "statistics'Test";
 SET citus.next_shard_id TO 980000;
 SET client_min_messages TO WARNING;
 SET citus.shard_count TO 32;
@@ -17,27 +17,51 @@ SELECT create_distributed_table('test_stats', 'a');
 CREATE STATISTICS s1 (dependencies) ON a, b FROM test_stats;
 
 -- test for distributing an already existing statistics
-CREATE TABLE test_stats2 (
+CREATE TABLE "test'stats2" (
     a   int,
     b   int
 );
 
-CREATE STATISTICS s2 (dependencies) ON a, b FROM test_stats2;
+CREATE STATISTICS s2 (dependencies) ON a, b FROM "test'stats2";
 
-SELECT create_distributed_table('test_stats2', 'a');
+SELECT create_distributed_table('test''stats2', 'a');
 
 -- test when stats is on a different schema
 CREATE SCHEMA sc1;
-CREATE TABLE tbl (a int, b int);
+CREATE TABLE tbl (a int, "B" text);
 SELECT create_distributed_table ('tbl', 'a');
 
-CREATE STATISTICS sc1.st1 ON a, b FROM tbl;
+CREATE STATISTICS sc1.st1 ON a, "B" FROM tbl;
+
+-- test distributing table with already created stats on a new schema
+CREATE TABLE test_stats3 (
+    a int,
+    b int
+);
+CREATE SCHEMA sc2;
+CREATE STATISTICS sc2."neW'Stat" ON a,b FROM test_stats3;
+SELECT create_distributed_table ('test_stats3','a');
 
 \c - - - :worker_1_port
-SELECT stxname FROM pg_statistic_ext ORDER BY stxname ASC;
+SELECT stxname
-SELECT count(DISTINCT stxnamespace) FROM pg_statistic_ext;
+FROM pg_statistic_ext
+WHERE stxnamespace IN (
+	SELECT oid
+	FROM pg_namespace
+	WHERE nspname IN ('public', 'statistics''Test', 'sc1', 'sc2')
+)
+ORDER BY stxname ASC;
+
+SELECT count(DISTINCT stxnamespace)
+FROM pg_statistic_ext
+WHERE stxnamespace IN (
+	SELECT oid
+	FROM pg_namespace
+	WHERE nspname IN ('public', 'statistics''Test', 'sc1', 'sc2')
+);
 
 \c - - - :master_port
 SET client_min_messages TO WARNING;
-DROP SCHEMA "statistics'test" CASCADE;
+DROP SCHEMA "statistics'Test" CASCADE;
 DROP SCHEMA sc1 CASCADE;
+DROP SCHEMA sc2 CASCADE;