Querry result not clear

Question

I have a function f1 and one table rno1 :->
create or replace function f1(p1 number ) return number is
temp1 number := 0 ;
begin
select s3.nextval into temp1 from dual;
return temp1;
end;
SQL> desc rno1;
Name Type
DEPTNO NUMBER(2)
RNO NUMBER(4)
Data in table rno1 is like :->
SQL> select DEPTNO from rno1;
DEPTNO
20
30
30
20
30
30
10
20
10
30
20
30
20
10
20
20
20
20
20
19 rows selected.
When I am executing the below querry i am getting for Deptno 10 two DEPTNO1 values
1184 and 1185 For the deptno 20 one value 1186 and for Deptno 30 two values
1187 and 1188.
1 select r1.DEPTNO, r2.deptno1 from rno1 r1,
2 (select DEPTNO, f1(c.deptno) DEPTNO1 from rno1 c group by c.deptno) r2
3* where r1.DEPTNO = r2.DEPTNO
SQL> /
DEPTNO DEPTNO1
10 1184
10 1185
10 1185
20 1186
20 1186
20 1186
20 1186
20 1186
20 1186
20 1186
20 1186
20 1186
20 1186
30 1187
30 1187
30 1187
30 1188
30 1188
30 1188
19 rows selected.
Can any one explain this result querry.
I think for each deptno the should be one value only.

Answer 1

Hello,
Not exactly sure why it is happening but from the tests I did, it seems that a sort join/merge join combination causes the additional calls to the function. When the query uses a hash join, the problem does not ocurr. This can't be taken as a rule of thumb however as I've only tested it with this limited scenario:
SQL> select * from v$version;
Oracle9i Enterprise Edition Release 9.2.0.4.0 - 64bit Production
PL/SQL Release 9.2.0.4.0 - Production
CORE    9.2.0.3.0       Production
TNS for Solaris: Version 9.2.0.4.0 - Production
NLSRTL Version 9.2.0.4.0 - Production
SQL> select * from dt_test_tab;
    DEPTNO
        20
        30
        30
        20
        30
        30
        10
        20
        10
        30
        20
        30
        20
        10
        20
        20
        20
        20
        20
--No stats gathered
SQL> select
  2     r1.DEPTNO,
  3     r2.deptno1
  4  from
  5     dt_test_tab r1,
  6     (select
  7             DEPTNO,
  8             dt_test_f1(c.deptno) DEPTNO1
  9     from
10             dt_test_tab c
11     group by
12             c.deptno
13     ) r2
14  where
15     r1.DEPTNO = r2.DEPTNO;
    DEPTNO    DEPTNO1
        10        116
        10        117
        10        117
        20        118
        20        118
        20        118
        20        118
        20        118
        20        118
        20        118
        20        118
        20        118
        20        118
        30        119
        30        119
        30        119
        30        120
        30        120
        30        120
PLAN_TABLE_OUTPUT
| Id  | Operation            |  Name        | Rows  | Bytes | Cost  |
|   0 | SELECT STATEMENT     |              |       |       |       |
|   1 |  MERGE JOIN          |              |       |       |       |
|   2 |   VIEW               |              |       |       |       |
|   3 |    SORT GROUP BY     |              |       |       |       |
|   4 |     TABLE ACCESS FULL| DT_TEST_TAB  |       |       |       |
|*  5 |   SORT JOIN          |              |       |       |       |
|   6 |    TABLE ACCESS FULL | DT_TEST_TAB  |       |       |       |
Predicate Information (identified by operation id):
   5 - access("R1"."DEPTNO"="R2"."DEPTNO")
       filter("R1"."DEPTNO"="R2"."DEPTNO")
Note: rule based optimization
--Results after gather_table_stats appear to be correct:
    DEPTNO    DEPTNO1
        20        122
30 123
30 123
        20        122
30 123
30 123
10 121
        20        122
10 121
30 123
        20        122
30 123
        20        122
10 121
        20        122
        20        122
        20        122
        20        122
        20        122
| Id  | Operation            |  Name        | Rows  | Bytes | Cost  |
|   0 | SELECT STATEMENT     |              |    19 |   551 |     7 |
|*  1 |  HASH JOIN           |              |    19 |   551 |     7 |
|   2 |   VIEW               |              |     3 |    78 |     4 |
|   3 |    SORT GROUP BY     |              |     3 |     9 |     4 |
|   4 |     TABLE ACCESS FULL| DT_TEST_TAB  |    19 |    57 |     2 |
|   5 |   TABLE ACCESS FULL  | DT_TEST_TAB  |    19 |    57 |     2 |
Predicate Information (identified by operation id):
   1 - access("R1"."DEPTNO"="R2"."DEPTNO")
--Results after adding the order by to get the above results in the right order
SQL> select
  2     r1.DEPTNO,
  3     r2.deptno1
  4  from
  5     dt_test_tab r1,
  6     (select
  7             DEPTNO,
  8             dt_test_f1(c.deptno) DEPTNO1
  9     from
10             dt_test_tab c
11     group by
12             c.deptno
13     ) r2
14  where
15     r1.DEPTNO = r2.DEPTNO
16 order by
17 r1.deptno;
    DEPTNO    DEPTNO1
        10        124
        10        125
        10        125
        20        126
        20        126
        20        126
        20        126
        20        126
        20        126
        20        126
        20        126
        20        126
        20        126
        30        127
        30        127
        30        127
        30        128
        30        128
        30        128
| Id  | Operation            |  Name        | Rows  | Bytes | Cost  |
|   0 | SELECT STATEMENT     |              |    19 |   361 |     8 |
|   1 |  MERGE JOIN          |              |    19 |   361 |     8 |
|   2 |   VIEW               |              |     3 |    48 |     4 |
|   3 |    SORT GROUP BY     |              |     3 |     9 |     4 |
|   4 |     TABLE ACCESS FULL| DT_TEST_TAB  |    19 |    57 |     2 |
|*  5 |   SORT JOIN          |              |    19 |    57 |     4 |
|   6 |    TABLE ACCESS FULL | DT_TEST_TAB  |    19 |    57 |     2 |
Predicate Information (identified by operation id):
   5 - access("R1"."DEPTNO"="R2"."DEPTNO")
       filter("R1"."DEPTNO"="R2"."DEPTNO")I think the point is though that this isn't an reliable method of ensuring one call per deptno as the results change with the execution plan...the reasons for which however are beyond my understanding, although I'm sure someone else will have a better idea as to why this is happening.
HTH
David