两者之间的差异：NOT IN（IS NOT NULL）VS IN（IS NULL）

Question

Im trying to write a query to list the id and names of departments that don't have any employee. I have found a query that works but I don't understand why my approach is not working.

Both queries are the same except for the subquery inside the WHERE clause.

My NOT WORKING approach: IN...IS NULL (EDIT: IT RETURNS NO VALUES)

SELECT
    department_id,
    department_name
FROM
    departments
WHERE
    department_id IN ( 
        SELECT
            department_id
        FROM
            employees
        WHERE
            department_id IS NULL
    )
ORDER BY
    department_name;

The WORKING approach: NOT IN... IS NOT NULL

SELECT
    department_id,
    department_name
FROM
    departments
WHERE
    department_id NOT IN ( --why -> IN... IS NULL is not working?
        SELECT
            department_id
        FROM
            employees
        WHERE
            department_id IS NOT NULL
    )
ORDER BY
    department_name;

Answer 1

NULL == NULL is never true. So you are asking if department_id is in a set of department_id s where department_id is NULL. This will NEVER be true. In the second query, you're looking for department_id s that are NOT in a list of department_id s where department_id is NOT NULL . This will give you both NULL department_id s AND department_id s that are not in the employees table.

I would note that your WHERE clause in the subquery is not needed:

WHERE
department_id NOT IN (
    SELECT department_id FROM employees
)

Answer 2

in your 1st query inside subquery

 SELECT  department_id
        FROM
            employees
        WHERE
            department_id IS NULL

It does not return any department_id thats why your output will be null

An in statement will do col=val1 or col=val2 or col=val3. Putting a null in there will boil down to col=null which won't work

as a result in case of in statement it is good filter out null

 SELECT  department_id
            FROM
                employees
            WHERE
                department_id IS not NULL --filter out null

In your 2nd query

SELECT
    department_id,
    department_name
FROM
    departments
WHERE
    department_id NOT IN ( --why -> IN... IS NULL is not working?
        SELECT
            department_id
        FROM
            employees
        WHERE
            department_id IS NOT NULL -- null checking is good 
    )
ORDER BY
    department_name;

in sub query you cheked WHERE department_id IS NOT NULL which safe you from col=null type execution and those department_id return in subquery will be filterout

Answer 3

I would use NOT EXISTS instead :

SELECT d.*
FROM departments d
WHERE NOT EXISTS (SELECT 1 FROM employees e WHERE e.department_id = d.department_id);

If the subquery return null then that expression considered as false .

So, value = NULL will evaluate as NULL or UNKNOWN , So, you can use NOT EXISTS instead.

Answer 4

You can use a LEFT JOIN with a WHERE ... IS NULL condition in the WHERE clause, as follows :

SELECT
    d.department_id,
    d.department_name
FROM
    departments d
    LEFT JOIN employees e 
        ON e.department_id = d.department_id
WHERE e.department_id IS NULL

Answer 5

If we forget about the is null for a moment, let's consider the logic of your first query:

SELECT
    department_id,
    department_name
FROM
    departments
WHERE
    department_id IN ( 
        SELECT
            department_id
        FROM
            employees
    )
ORDER BY
    department_name;

This returns the set of departments which have employees. By definition, the result is bounded by the set of employees.department_id . Re-introducing the where department_id is null to the subquery constrains that set to null or produces an empty set.

Simply, there is no way we can conjure up the set of values which aren't in a table by querying only that table. Consequently we have to use the NOT IN (or NOT EXISTS if the table in the subquery contains null entries) or the OUTER JOIN solution suggested by others.

Answer 6

This expression: X IN ( a,b,c ) is equivalent to: X = a OR X = b OR X = c

This expession: NOT X IN ( a,b,c ) is equivalent to: NOT (X = a OR X = b OR X = c) , which in turn is eqivalent to: NOT X = a AND NOT X = b AND NOT X = c

You also need to study Comparisons with NULL and the three-valued logic (3VL)

If you know the above, you can create a truth table for each of expressions and for for each combination of values - this will help you to gain understanding of behaviour of IN/NOT IN in SQL Queries:

+------+------+---+---+--------------+------------------+
|  X   |  a   | b | c | X in (a,b,c) | NOT X in (a,b,c) |
+------+------+---+---+--------------+------------------+
| 1    | 0    | 1 | 2 | true         | false            |
| 1    | NULL | 1 | 2 | false        | NULL(false) *    |
| 1    | 0    | 2 | 3 | false        | true             |
| 1    | NULL | 2 | 3 | false        | NULL(false) *    |
| NULL | 0    | 1 | 2 | NULL(false)  | NULL(false) *    |
| NULL | NULL | 1 | 2 | NULL(false)  | NULL(false) *    |
+------+------+---+---+--------------+------------------+

Please pay a special attenton for rows marked with * - these are the ones that are not consistent with common sense where if X = false then NOT X must be true

Answer 7

What you need here is: WHERE NOT EXISTS since what you need is all departments where no employee exists.

select department_id, department_name from departments 
where not exists (
  select 1 from employees where employees.department_id = departments.department_id
)

using where is null/is not null is wrong in both cases here.