# PostgreSQL: Can i use tuple in where clause?

Actually might be a simple question:

Given 2 tables:

``````table1: A,B,C
(A1,B1,28471),
(A1,B2,01244),
(A2,B1,1283a),
(A2,B2,82r7e);

table2: A,B,D,E,F,G
(A1,B1,18,1,6,8),
(A2,B2,18,2,3,0),
(A3,B1,18,7,1,4),
(A4,B2,18,1,9,6);
``````

Will the following statement result in a the following result given the example tables:

``````SELECT E,F,G FROM table2 WHERE (A,B) IN (SELECT A,B FROM table1)
``````

expected result:

`````` E F G
(1,6,8), -- (A1,B1)
(2,3,0); -- (A2,B2)
``````

Can someone confirm this is a valid method to select based on tuples?

asking as the most likely next step is to UPDATE the E field by adding +5 for matching results

Consider `EXISTS` instead:

``````SELECT E,F,G
FROM   table2 t2
WHERE  EXISTS (
SELECT FROM table1 t1
WHERE  (t1.A,t1.B) = (t2.A,t2.B)
);
``````

db<>fiddle here

Equivalent and typically cheaper.
And the negation `NOT EXISTS` is not as treacherous as the discouraged (not completely equivalent) `NOT IN`. See:

Aside, I would call `(A,B)` a "composite type" or "row type" rather than a "tuple".

• In fact, `(a,b,c)` is another syntax for `ROW(a,b,c)` - `ROW` keyword is optional, docs Commented Jun 13, 2023 at 4:47

Thanks to Peter Vandivier (couldnt flag his comment as the answer)

Yes it is possible:

What have you tried😉 – Peter Vandivier 12 mins ago

• To be fair, I literally just copy-pasted your question onto db<>fiddle, so it feels like a self-answer is appropriate here 🤪 Commented Apr 29, 2021 at 8:40
• Not so important here, but table aliases are a good idea? Commented Apr 29, 2021 at 10:02