From 38baddbc4197f87d14db70016a3bd6947bfaf523 Mon Sep 17 00:00:00 2001
From: Rock <yangzhizheng@chinamobile.com>
Date: Tue, 11 Jun 2024 17:27:02 +0800
Subject: [PATCH] Fix the sorting error issue for the set type.

Conflicts:
	src/test/regress/expected/mysql_syntax.out
	src/test/regress/sql/mysql_syntax.sql
---
 src/gausskernel/optimizer/plan/initsplan.cpp  | 18 ++++++++++++-
 src/test/regress/expected/hw_datatype_set.out | 10 +++----
 src/test/regress/expected/mysql_syntax.out    | 26 +++++++++++++++++++
 src/test/regress/sql/hw_datatype_set.sql      |  4 +--
 src/test/regress/sql/mysql_syntax.sql         | 11 ++++++++
 5 files changed, 61 insertions(+), 8 deletions(-)

diff --git a/src/gausskernel/optimizer/plan/initsplan.cpp b/src/gausskernel/optimizer/plan/initsplan.cpp
index 8ce63ff77..d99fd5b8d 100644
--- a/src/gausskernel/optimizer/plan/initsplan.cpp
+++ b/src/gausskernel/optimizer/plan/initsplan.cpp
@@ -2105,7 +2105,7 @@ static void check_mergejoinable(RestrictInfo* restrictinfo)
     Expr* clause = restrictinfo->clause;
     Oid opno;
     Node* leftarg = NULL;
-
+    Node* rightarg = NULL;
     if (restrictinfo->pseudoconstant)
         return;
     if (!is_opclause(clause))
@@ -2115,6 +2115,22 @@ static void check_mergejoinable(RestrictInfo* restrictinfo)
 
     opno = ((OpExpr*)clause)->opno;
     leftarg = (Node*)linitial(((OpExpr*)clause)->args);
+    rightarg = (Node*)lsecond(((OpExpr*)clause)->args);
+    /* 
+     * For data of type set, using the merge join plan will result in inconsistent outcomes.
+     */
+    if (nodeTag(leftarg) == T_Var) {
+        Var* leftCol = (Var*)leftarg;
+        if (type_is_set(leftCol->vartype)) {
+            return;
+        }
+    }
+    if (nodeTag(rightarg) == T_Var) {
+        Var* rightCol = (Var*)rightarg;
+        if (type_is_set(rightCol->vartype)) {
+            return;
+        }     
+    }
     if (op_mergejoinable(opno, exprType(leftarg)) && !contain_volatile_functions((Node*)clause))
         restrictinfo->mergeopfamilies = get_mergejoin_opfamilies(opno);
 
diff --git a/src/test/regress/expected/hw_datatype_set.out b/src/test/regress/expected/hw_datatype_set.out
index 556ad6000..ef43f477d 100644
--- a/src/test/regress/expected/hw_datatype_set.out
+++ b/src/test/regress/expected/hw_datatype_set.out
@@ -4428,13 +4428,13 @@ select
 from
   test_collation2 as t2,
   test_collation2 as t3
-where t2.a = t3.a;
+where t2.a = t3.a order by 1;
    a    |   a    
 --------+--------
- aaa    | aaa
- 高斯ss | 高斯ss
- 高斯SS | 高斯SS
  AAA    | AAA
+ aaa    | aaa
+ 高斯SS | 高斯SS
+ 高斯ss | 高斯ss
 (4 rows)
 
 -- '<>'
@@ -5344,4 +5344,4 @@ DETAIL:  N/A
 insert into test_check1 values('B  ');
 insert into test_check1 values('C');
 insert into test_check1 values('c');
-drop table test_check1;
\ No newline at end of file
+drop table test_check1;
diff --git a/src/test/regress/expected/mysql_syntax.out b/src/test/regress/expected/mysql_syntax.out
index d9d7d5bb6..11c5f9170 100644
--- a/src/test/regress/expected/mysql_syntax.out
+++ b/src/test/regress/expected/mysql_syntax.out
@@ -526,6 +526,32 @@ NOTICE:  SQLSTATE = 22012, SQLERRM = division by zero
  
 (1 row)
 
+drop table if exists test1;
+NOTICE:  table "test1" does not exist, skipping
+drop table if exists test2;
+NOTICE:  table "test2" does not exist, skipping
+create table test1 (c set('a','b','c','d'));
+NOTICE:  CREATE TABLE will create implicit set "test1_c_set" for column "test1.c"
+insert into test1 values ('a'),('b'),('c'),('d');
+create table test2 (c set('d','c','b','a'));
+NOTICE:  CREATE TABLE will create implicit set "test2_c_set" for column "test2.c"
+insert into test2 values ('a'),('b'),('c'),('d');
+set enable_nestloop=off;
+set enable_hashjoin=off;
+explain (costs off) select test1.c from test1, test2 where test1.c = test2.c order by 1;
+                QUERY PLAN                
+------------------------------------------
+ Sort
+   Sort Key: test1.c
+   ->  Nested Loop
+         Join Filter: (test1.c = test2.c)
+         ->  Seq Scan on test1
+         ->  Materialize
+               ->  Seq Scan on test2
+(7 rows)
+
+set enable_nestloop=on;
+set enable_hashjoin=on;
 \c regression
 drop trigger animal_trigger1;
 ERROR:  drop trigger without table name only support in B-format database
diff --git a/src/test/regress/sql/hw_datatype_set.sql b/src/test/regress/sql/hw_datatype_set.sql
index fd4bebb5e..a8848346b 100644
--- a/src/test/regress/sql/hw_datatype_set.sql
+++ b/src/test/regress/sql/hw_datatype_set.sql
@@ -879,7 +879,7 @@ select
 from
   test_collation2 as t2,
   test_collation2 as t3
-where t2.a = t3.a;
+where t2.a = t3.a order by 1;
 
 -- '<>'
 select a, b from test_collation_op where a <> b; -- set <> other set
@@ -1118,4 +1118,4 @@ insert into test_check1 values('a');
 insert into test_check1 values('B  ');
 insert into test_check1 values('C');
 insert into test_check1 values('c');
-drop table test_check1;
\ No newline at end of file
+drop table test_check1;
diff --git a/src/test/regress/sql/mysql_syntax.sql b/src/test/regress/sql/mysql_syntax.sql
index 0bf175198..bdf12f1d9 100644
--- a/src/test/regress/sql/mysql_syntax.sql
+++ b/src/test/regress/sql/mysql_syntax.sql
@@ -411,6 +411,17 @@ exception
 END;
 /
 call test_condition_7();
+drop table if exists test1;
+drop table if exists test2;
+create table test1 (c set('a','b','c','d'));
+insert into test1 values ('a'),('b'),('c'),('d');
+create table test2 (c set('d','c','b','a'));
+insert into test2 values ('a'),('b'),('c'),('d');
+set enable_nestloop=off;
+set enable_hashjoin=off;
+explain (costs off) select test1.c from test1, test2 where test1.c = test2.c order by 1;
+set enable_nestloop=on;
+set enable_hashjoin=on;
 \c regression
 drop trigger animal_trigger1;
 drop trigger if exists animal_trigger1;