Trouble updating multiple rows in table using subquery

Question

Hi everyone, I'm having trouble updating multiple rows with a subquery. Here's the setup:
create table mytable (
col_a number primary key,
col_b number,
col_c number,
col_d number);
insert into mytable values (1 ,1,1,15);
insert into mytable values (2 ,1,2,7 );
insert into mytable values (3 ,1,3,11);
insert into mytable values (4 ,1,4,23);
insert into mytable values (5 ,1,5,14);
insert into mytable values (6 ,2,1,50);
insert into mytable values (7 ,2,2,41);
insert into mytable values (8 ,2,3,13);
insert into mytable values (9 ,2,4,12);
insert into mytable values (10,2,5,19);
insert into mytable values (11,3,1,10);
insert into mytable values (12,3,2,92);
insert into mytable values (13,3,3,81);
insert into mytable values (14,3,4,17);
insert into mytable values (15,3,5,66);
insert into mytable values (16,4,1,54);
insert into mytable values (17,4,2,41);
insert into mytable values (18,4,3,22);
insert into mytable values (19,4,4,24);
insert into mytable values (20,4,5,17);For this example, using an update statement (or merge if that's better), say I want to set the values for col_d where col_b = 3 equal to the values for col_d where col_b = 1 and col_c equal each other. Results should look like the following after the update:
col_a col_b col_c col_d
1     1     1     15
2     1     2     7
3     1     3     11
4     1     4     23
5     1     5     14
6     2     1     50
7     2     2     41
8     2     3     13
9     2     4     12
10    2     5     19
11    3     1     15
12    3     2     7
13    3     3     11
14    3     4     23
15    3     5     14
16    4     1     54
17    4     2     41
18    4     3     22
19    4     4     24
20    4     5     17I can see it right there at my fingertips using this query, where I want to set b_col_d = a_col_d, but I'm missing something, as this query returns too many rows when used in the update statement.
select * from (
  select col_a as a_col_a, col_b as a_col_b, col_c as a_col_c, col_d as a_col_d
  from mytable
  where col_b = 1
  ) a, (
  select col_a as b_col_a, col_b as b_col_b, col_c as b_col_c, col_d as b_col_d
  from mytable
  where col_b = 3
  ) b
where a.a_col_c = b.b_col_cupdate mytable set column_d = (select ??? where exists ???)
Can someone help me get there? I'm using 10GR2.
Thanks!
Mark

Answer 1

Hopefully this is what you are looking for:
SQL > UPDATE mytable myt1
  2  SET    col_d = ( SELECT myt2.col_d
  3                   FROM   mytable myt2
  4                   WHERE  myt2.col_b = 1
  5                   AND    myt1.col_c = myt2.col_c
  6                 )
  7  WHERE  col_b = 3
  8  AND    EXISTS
  9         ( SELECT NULL
10           FROM   mytable myt2
11           WHERE  myt2.col_c = myt1.col_c
12         )
13  ;
5 rows updated.
SQL > SELECT * FROM mytable ORDER BY col_a;
               COL_A                COL_B                COL_C                COL_D
                   1                    1                    1                   15
                   2                    1                    2                    7
                   3                    1                    3                   11
                   4                    1                    4                   23
                   5                    1                    5                   14
                   6                    2                    1                   50
                   7                    2                    2                   41
                   8                    2                    3                   13
                   9                    2                    4                   12
                  10                    2                    5                   19
                  11                    3                    1                   15
                  12                    3                    2                    7
                  13                    3                    3                   11
                  14                    3                    4                   23
                  15                    3                    5                   14
                  16                    4                    1                   54
                  17                    4                    2                   41
                  18                    4                    3                   22
                  19                    4                    4                   24
                  20                    4                    5                   17
20 rows selected.Thank you so much for providing the sample data in an easy to consume form, as well as the expected output.