[MCOL-4699] Add support for circular outer joins.

dbcon/joblist/jlf_common.h

@@ -28,6 +28,7 @@
 #include <stack>
 #include <string>
 #include <vector>
+#include <unordered_map>
 
 #include <boost/shared_ptr.hpp>
 #include <boost/uuid/uuid.hpp>
@@ -365,6 +366,10 @@ struct JobInfo
   std::vector<execplan::ParseTree*> dynamicParseTreeVec;
 
   PrimitiveServerThreadPools primitiveServerThreadPools;
+  // Represents a `join edges` and `join id` to be restored in `join order` part.
+  std::map<std::pair<uint32_t, uint32_t>, uint32_t> joinEdgesToRestore;
+  // Represents a pair of `table` to be on a large side and weight associated with that table.
+  std::unordered_map<uint32_t, uint32_t> tablesForLargeSide;
 
  private:
   // defaults okay

dbcon/joblist/jlf_tuplejoblist.h

@@ -128,19 +128,31 @@ void orExpresssion(const execplan::Operator* op, JobInfo& jobInfo);
 // union the queries and return the tuple union step
 SJSTEP unionQueries(JobStepVector& queries, uint64_t distinctUnionNum, JobInfo& jobInfo);
 
+// Used for join graph analysis.
+// WHITE - node is not processed.
+// GREY - node is in process.
+// BLACK - node is done.
+enum class JoinTableColor
+{
+  WHITE,
+  GREY,
+  BLACK
+};
+
 struct JoinTableNode
 {
-  bool fVisited;
+  JoinTableColor fTableColor;
   uint32_t fParent;
   std::vector<uint32_t> fAdjacentList;
-  JoinTableNode() : fVisited(false), fParent(-1)
+  JoinTableNode() : fTableColor(JoinTableColor::WHITE), fParent(UINT_MAX)
   {
   }
 };
 
 using JoinGraph = std::map<uint32_t, JoinTableNode>;
-using JoinEdges = std::set<pair<uint32_t, uint32_t>>;
+using JoinEdge = std::pair<uint32_t, uint32_t>;
-using Cycles = std::vector<std::vector<std::pair<uint32_t, uint32_t>>>;
+using JoinEdges = std::set<JoinEdge>;
-using Cycle = std::vector<std::pair<uint32_t, uint32_t>>;
+using Cycle = std::vector<JoinEdge>;
-using PostJoinFilterKeys = std::vector<std::pair<std::pair<uint32_t, uint32_t>, std::vector<uint32_t>>>;
+using Cycles = std::vector<std::vector<JoinEdge>>;
+using PostJoinFilterKeys = std::vector<std::pair<JoinEdge, std::vector<uint32_t>>>;
 }  // namespace joblist

dbcon/joblist/tuplehashjoin.cpp

@@ -1092,7 +1092,8 @@ const string TupleHashJoinStep::toString() const
 {
   ostringstream oss;
   size_t idlsz = fInputJobStepAssociation.outSize();
-  idbassert(idlsz > 1);
+  // Avoid assertion on empty `TupleHashJoinStep`.
+  idbassert(idlsz > 1 || idlsz == 0);
   oss << "TupleHashJoinStep    ses:" << fSessionId << " st:" << fStepId;
   oss << omitOidInDL;
 
@@ -1184,6 +1185,7 @@ void TupleHashJoinStep::configJoinKeyIndex(const vector<JoinType>& jt, const vec
 {
   joinTypes.insert(joinTypes.begin(), jt.begin(), jt.end());
   typelessJoin.insert(typelessJoin.begin(), typeless.begin(), typeless.end());
+
   smallSideKeys.insert(smallSideKeys.begin(), smallkey.begin(), smallkey.end());
   largeSideKeys.insert(largeSideKeys.begin(), largekey.begin(), largekey.end());
 #ifdef JLF_DEBUG

mysql-test/columnstore/basic/r/mcol-4699.result

@@ -0,0 +1,133 @@
+DROP DATABASE IF EXISTS mcol4699;
+CREATE DATABASE mcol4699;
+USE mcol4699;
+create table t1 (a int, b int) engine=columnstore;
+create table t2 (a int, b int) engine=columnstore;
+create table t3 (a int, b int) engine=columnstore;
+create table t4 (a int, b int) engine=columnstore;
+create table t5 (a int, b int) engine=columnstore;
+create table t6 (a int, b int) engine=columnstore;
+create table t7 (a int, b int) engine=columnstore;
+create table t8 (a int, b int) engine=columnstore;
+create table t9 (a int, b int) engine=columnstore;
+insert into t1 values (1, 3), (2, 3), (3, 4);
+insert into t2 values (1, 2), (2, 4), (4, 5);
+insert into t3 values (1, 2), (2, 3), (3, 4), (4, 5);
+insert into t4 values (1, 3);
+insert into t5 values (1, 2), (3, 4);
+insert into t6 values (1, 2), (3, 4);
+insert into t7 values (1, 3);
+insert into t8 values (1, 3);
+insert into t9 values (1, 2);
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b);
+a	b	a	b	a	b
+1	3	1	2	1	2
+NULL	NULL	NULL	NULL	2	3
+NULL	NULL	NULL	NULL	3	4
+NULL	NULL	NULL	NULL	4	5
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) left join t4 on (t3.a = t4.a);
+a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3
+NULL	NULL	NULL	NULL	2	3	NULL	NULL
+NULL	NULL	NULL	NULL	3	4	NULL	NULL
+NULL	NULL	NULL	NULL	4	5	NULL	NULL
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) where (t3.a = 1 or t3.a = 3);
+a	b	a	b	a	b
+1	3	1	2	1	2
+NULL	NULL	NULL	NULL	3	4
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) where (t3.a > 1);
+a	b	a	b	a	b
+NULL	NULL	NULL	NULL	2	3
+NULL	NULL	NULL	NULL	3	4
+NULL	NULL	NULL	NULL	4	5
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t2.a = t3.a) left join t4 on (t3.a = t4.a and t1.b = t4.b) order by t3.a;
+a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3
+2	3	2	4	2	3	NULL	NULL
+NULL	NULL	NULL	NULL	3	4	NULL	NULL
+NULL	NULL	NULL	NULL	4	5	NULL	NULL
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t2.a = t3.a) left join t4 on (t3.a = t4.a and t2.b = t4.b) order by t3.a;
+a	b	a	b	a	b	a	b
+1	3	1	2	1	2	NULL	NULL
+2	3	2	4	2	3	NULL	NULL
+NULL	NULL	NULL	NULL	3	4	NULL	NULL
+NULL	NULL	NULL	NULL	4	5	NULL	NULL
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b and t3.a > t1.a);
+a	b	a	b	a	b
+NULL	NULL	NULL	NULL	1	2
+NULL	NULL	NULL	NULL	2	3
+NULL	NULL	NULL	NULL	3	4
+NULL	NULL	NULL	NULL	4	5
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b and t3.a > t2.a and t3.a > t1.a);
+a	b	a	b	a	b
+NULL	NULL	NULL	NULL	1	2
+NULL	NULL	NULL	NULL	2	3
+NULL	NULL	NULL	NULL	3	4
+NULL	NULL	NULL	NULL	4	5
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b and t3.b > t2.a and t3.a < t1.b);
+a	b	a	b	a	b
+1	3	1	2	1	2
+NULL	NULL	NULL	NULL	2	3
+NULL	NULL	NULL	NULL	3	4
+NULL	NULL	NULL	NULL	4	5
+select * from t1 left join t2 on (t1.a = t2.a) left join t3 on (t2.a = t3.a) left join t4 on (t4.a = t3.a) left join t5 on (t5.a = t2.a) left join t6 on (t5.a = t6.a and t6.a = t4.a);
+a	b	a	b	a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3	1	2	1	2
+2	3	2	4	2	3	NULL	NULL	NULL	NULL	NULL	NULL
+3	4	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) left join t4 on (t3.a = t4.a and t2.a = t4.a) order by t3.a;
+a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3
+NULL	NULL	NULL	NULL	2	3	NULL	NULL
+NULL	NULL	NULL	NULL	3	4	NULL	NULL
+NULL	NULL	NULL	NULL	4	5	NULL	NULL
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) right join t4 on (t3.a = t4.a and t2.a = t4.a) order by t3.a;
+a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a) left join t4 on (t3.a = t4.a and t1.a = t4.a and t2.a = t4.a);
+a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3
+2	3	2	4	2	3	NULL	NULL
+NULL	NULL	NULL	NULL	3	4	NULL	NULL
+NULL	NULL	NULL	NULL	4	5	NULL	NULL
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a) right join t4 on (t3.a = t4.a and t1.a = t4.a and t2.a = t4.a);
+a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3
+select * from t1 left join t2 on (t1.a = t2.a) left join t3 on (t2.a = t3.a) left join t4 on (t4.a = t3.a) left join t5 on (t5.a = t2.a) left join t6 on (t5.a = t6.a and t6.a = t4.a) left join t7 on (t7.a = t3.a) left join t8 on (t8.a = t7.a and t8.a = t2.a);
+a	b	a	b	a	b	a	b	a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3	1	2	1	2	1	3	1	3
+2	3	2	4	2	3	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL
+3	4	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) left join t4 on (t3.a = t4.a and t2.a = t4.a) left join t5 on (t4.a = t5.a and t3.b = t5.b);
+a	b	a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3	1	2
+NULL	NULL	NULL	NULL	2	3	NULL	NULL	NULL	NULL
+NULL	NULL	NULL	NULL	3	4	NULL	NULL	NULL	NULL
+NULL	NULL	NULL	NULL	4	5	NULL	NULL	NULL	NULL
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) left join t4 on (t3.a = t4.a and t2.a = t4.a) right join t5 on (t4.a = t5.a and t3.b = t5.b);
+a	b	a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3	1	2
+NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	3	4
+select * from t1 left join t2 on (t1.a = t2.a) left join t3 on (t2.a = t3.a) left join t4 on (t4.a = t3.a) left join t5 on (t5.a = t2.a) left join t6 on (t5.a = t6.a and t6.a = t4.a) left join t7 on (t7.a = t3.a) left join t8 on (t8.a = t7.a and t8.a = t2.a) left join t9 on (t7.a = t9.a and t4.a = t9.a);
+a	b	a	b	a	b	a	b	a	b	a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3	1	2	1	2	1	3	1	3	1	2
+2	3	2	4	2	3	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL
+3	4	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) left join t4 on (t3.a = t4.a and t2.a = t4.a) right join t5 on (t4.a = t5.a and t3.b = t5.b) right join t6 on (t5.a = t6.a and t4.a = t6.a);
+a	b	a	b	a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3	1	2	1	2
+NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	3	4
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) left join t4 on (t3.a = t4.a and t2.a = t4.a) right join t5 on (t4.a = t5.a and t3.b = t5.b) left join t6 on (t5.a = t6.a and t4.a = t6.a);
+a	b	a	b	a	b	a	b	a	b	a	b
+1	3	1	2	1	2	1	3	1	2	1	2
+NULL	NULL	NULL	NULL	NULL	NULL	NULL	NULL	3	4	NULL	NULL
+drop table t1;
+drop table t2;
+drop table t3;
+drop table t4;
+drop table t5;
+drop table t6;
+drop table t7;
+drop table t8;
+drop table t9;
+DROP DATABASE mcol4699;

mysql-test/columnstore/basic/t/mcol-4699.test

@@ -0,0 +1,70 @@
+
+#-- source ../include/have_columnstore.inc
+
+--disable_warnings
+DROP DATABASE IF EXISTS mcol4699;
+--enable_warnings
+
+CREATE DATABASE mcol4699;
+
+USE mcol4699;
+
+create table t1 (a int, b int) engine=columnstore;
+create table t2 (a int, b int) engine=columnstore;
+create table t3 (a int, b int) engine=columnstore;
+create table t4 (a int, b int) engine=columnstore;
+create table t5 (a int, b int) engine=columnstore;
+create table t6 (a int, b int) engine=columnstore;
+create table t7 (a int, b int) engine=columnstore;
+create table t8 (a int, b int) engine=columnstore;
+create table t9 (a int, b int) engine=columnstore;
+
+insert into t1 values (1, 3), (2, 3), (3, 4);
+insert into t2 values (1, 2), (2, 4), (4, 5);
+insert into t3 values (1, 2), (2, 3), (3, 4), (4, 5);
+insert into t4 values (1, 3);
+insert into t5 values (1, 2), (3, 4);
+insert into t6 values (1, 2), (3, 4);
+insert into t7 values (1, 3);
+insert into t8 values (1, 3);
+insert into t9 values (1, 2);
+
+# 1 cycle.
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b);
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) left join t4 on (t3.a = t4.a);
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) where (t3.a = 1 or t3.a = 3);
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) where (t3.a > 1);
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t2.a = t3.a) left join t4 on (t3.a = t4.a and t1.b = t4.b) order by t3.a;
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t2.a = t3.a) left join t4 on (t3.a = t4.a and t2.b = t4.b) order by t3.a;
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b and t3.a > t1.a);
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b and t3.a > t2.a and t3.a > t1.a);
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b and t3.b > t2.a and t3.a < t1.b);
+select * from t1 left join t2 on (t1.a = t2.a) left join t3 on (t2.a = t3.a) left join t4 on (t4.a = t3.a) left join t5 on (t5.a = t2.a) left join t6 on (t5.a = t6.a and t6.a = t4.a);
+
+# 2 cycles.
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) left join t4 on (t3.a = t4.a and t2.a = t4.a) order by t3.a;
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) right join t4 on (t3.a = t4.a and t2.a = t4.a) order by t3.a;
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a) left join t4 on (t3.a = t4.a and t1.a = t4.a and t2.a = t4.a);
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a) right join t4 on (t3.a = t4.a and t1.a = t4.a and t2.a = t4.a);
+select * from t1 left join t2 on (t1.a = t2.a) left join t3 on (t2.a = t3.a) left join t4 on (t4.a = t3.a) left join t5 on (t5.a = t2.a) left join t6 on (t5.a = t6.a and t6.a = t4.a) left join t7 on (t7.a = t3.a) left join t8 on (t8.a = t7.a and t8.a = t2.a);
+
+# 3 cycles.
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) left join t4 on (t3.a = t4.a and t2.a = t4.a) left join t5 on (t4.a = t5.a and t3.b = t5.b);
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) left join t4 on (t3.a = t4.a and t2.a = t4.a) right join t5 on (t4.a = t5.a and t3.b = t5.b);
+select * from t1 left join t2 on (t1.a = t2.a) left join t3 on (t2.a = t3.a) left join t4 on (t4.a = t3.a) left join t5 on (t5.a = t2.a) left join t6 on (t5.a = t6.a and t6.a = t4.a) left join t7 on (t7.a = t3.a) left join t8 on (t8.a = t7.a and t8.a = t2.a) left join t9 on (t7.a = t9.a and t4.a = t9.a);
+
+# 4 cycles.
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) left join t4 on (t3.a = t4.a and t2.a = t4.a) right join t5 on (t4.a = t5.a and t3.b = t5.b) right join t6 on (t5.a = t6.a and t4.a = t6.a);
+select * from t1 inner join t2 on (t1.a = t2.a) right join t3 on (t1.a = t3.a and t2.b = t3.b) left join t4 on (t3.a = t4.a and t2.a = t4.a) right join t5 on (t4.a = t5.a and t3.b = t5.b) left join t6 on (t5.a = t6.a and t4.a = t6.a);
+
+drop table t1;
+drop table t2;
+drop table t3;
+drop table t4;
+drop table t5;
+drop table t6;
+drop table t7;
+drop table t8;
+drop table t9;
+
+DROP DATABASE mcol4699;

utils/loggingcpp/ErrorMessage.txt

@@ -22,7 +22,6 @@
 1000	ERR_MISS_JOIN	%1% not joined.
 1001	ERR_NON_SUPPORTED_FUNCTION	Function '%1%' isn't supported.
 1002	ERR_INCOMPATIBLE_JOIN	%1% incompatible column type specified for join condition.
-1003	ERR_CIRCULAR_JOIN	Circular joins are not supported.
 1004	ERR_MIX_JOIN	Mixed %1% JOIN is not supported.
 1005	ERR_UPDATE_SUB	update with subselect in select clause is currently not supported in Columnstore.
 1006	ERR_DATATYPE_NOT_SUPPORT	Function called with unsupported datatype.