Prevent Oracle minus statement from removing duplicates

Question

Given these two tables:

CREATE TABLE TEST1 (TEST  VARCHAR2(1 BYTE))
CREATE TABLE TEST2 (TEST  VARCHAR2(1 BYTE))

Where TEST1 has two rows both w

Answer 1

SELECT TEST FROM TEST1 WHERE TEST IN(
SELECT TEST FROM TEST1
MINUS
SELECT TEST FROM TEST2);

Answer 2

The "NOT IN" answers are all correct. An alternative, which might be easier for some scenarios, is the "NOT EXISTS" operator:

SELECT TEST FROM TEST1
WHERE NOT EXISTS
(SELECT null FROM TEST2 WHERE TEST2.TEST = TEST1.TEST);

(Note: the "null" in the select clause is meaningless here)

I personally use both methods, but I like the NOT EXISTS often because it is more flexible - it doesn't require the comparison to be on an equality condition, for example.

Recent versions of the optimiser will often convert a NOT IN to a NOT EXISTS, or vice versa; however, if you're on an older version (e.g. 8i or even 9i I think) you may see performance benefits from switching between these two methods.

Answer 3

Part of what minus does is removes duplicates. Consider using NOT IN to avoid removing duplicates.

SELECT TEST FROM TEST1 WHERE TEST NOT IN(SELECT TEST FROM TEST2)

Answer 4

Oracle supports multiple columns in the IN statement, so you can write:

SELECT a, b, c
FROM table1
WHERE (a,b,c) not in (
    select a,b,c from table2
)

Answer 5

Another option:

SELECT TEST, row_number() OVER (PARTITION BY TEST ORDER BY TEST) FROM TEST1
MINUS
SELECT TEST, row_number() OVER (PARTITION BY TEST ORDER BY TEST) FROM TEST2

This would be MINUS with each duplicate treated as a distinct entry. Note in the example below, if TEST1 has two 'C' values and TEST2 has only one, you get one in the output.

dev> select * from test1;

T
-
A
A
B
C
C

dev> select * from test2;

T
-
B
C

dev>     SELECT TEST, row_number() OVER (PARTITION BY TEST ORDER BY TEST) FROM TEST1
  2      MINUS
  3      SELECT TEST, row_number() OVER (PARTITION BY TEST ORDER BY TEST) FROM TEST2
  4  /

T ROW_NUMBER()OVER(PARTITIONBYTESTORDERBYTEST)
- --------------------------------------------
A                                            1
A                                            2
C                                            2

Answer 6

SELECT field1 FROM table1 WHERE field1 NOT IN (SELECT field2 FROM table2)

Will work so long as field2 can't contain NULLs.