Using IN for a subquery - HR schema

Question

This relates to another post where I wanted to write a query from the HR schema that gave the the names and salaries of those employees (from the employees table) who earn the highest salaries in their respective departments. See Re: Using a subquery to display names and max salaries by department: HR Schema .
One of the replies to that post suggested the query:
SELECT first_name, last_name, salary
FROM employees
WHERE (department_id,salary) IN
(SELECT department_id, MAX(salary) FROM employees GROUP BY department_id);
This query works fine. However I still don't understand how the "in" works in this query.
From what I understand "in" is for a set of OR conditions?
Also if I change the order of the columns in the WHERE clause I get an error where no rows are fetched:
SELECT first_name, last_name, salary
FROM employees
WHERE (salary, department_id) IN (select department_id, max(salary) from employees group by department_id);
0 rows selected
If anyone could explain this to me it would be much appreciated.
I'm using Oracle Database 11g express edition and sql developer.

Answer 1

Hi,
993451 wrote:
This relates to another post where I wanted to write a query from the HR schema that gave the the names and salaries of those employees (from the employees table) who earn the highest salaries in their respective departments. See Re: Using a subquery to display names and max salaries by department: HR Schema .
One of the replies to that post suggested the query:
SELECT first_name, last_name, salary
FROM employees
WHERE (department_id,salary) IN
(SELECT department_id, MAX(salary) FROM employees GROUP BY department_id);
This query works fine. However I still don't understand how the "in" works in this query.
From what I understand "in" is for a set of OR conditions?Yes, that's baiscally right.
For example, the scott.dept table has 4 rows, with deptnos 10, 20, 30 and 40, so
WHERE   x  IN  (
                   SELECT  deptno
                   FROM    scott.dept
               )is equivalent to
WHERE   x  = 10
OR      x  = 20
OR      x  = 30
OR      x  = 40 Also if I change the order of the columns in the WHERE clause I get an error where no rows are fetched:
SELECT first_name, last_name, salary
FROM employees
WHERE (salary, department_id) IN (select department_id, max(salary) from employees group by department_id);
0 rows selectedThat's actually not an error; it's just not the results you wanted.
The order of the columns is significant. The N-th value in the list before the IN operator must match the N-th value in the SELECT clause of the sub-query.
When you say
WHERE   (x, y)  IN (
                       SELECT   a, b
                       FROM    ...
                   )That's saying "only include rows where x and y *both* match a and b, respectively, on the same row"
If anyone could explain this to me it would be much appreciated.Let's look at a concret example.
One of the rows produced by the sub-query <tt> (select department_id, max(salary) from employees group by department_id) </tt> has department_id=10 and max(salary) = 4400. That means there is a row in the table that has department_id=10 and salary=4400. But what happens when you evaluate the WHERE clause
WHERE (salary, department_id) IN (select department_id, max(salary) from employees group by department_id)for that row? that's equivalent to saying
WHERE   salary         = 10
AND     department_id  = 4400Both parts of that condition are FALSE, so the condition as a whole is FALSE.
I'm using Oracle Database 11g express edition and sql developer.Thanks for including that! In this case, it happens not to matter (that is, the behavior you asked about is the same in all recent versions of Oracle, and it does not depend on your front end), but you had no way of knowing that, it's easy to give your version, and it never hurts.