How does the NULL in SQL works in the below scenario?

Question

I am practicing SQL and realized that the NULL in the SQL are pretty amazing " entities ".

Table " xo " Please see the link for table structure and contents

When I run the following SQL query

select a from xo where a not in (select c from xo) ;

it returns no rows . When I run this query

select a from xo where a  in (select c from xo);

It works the way I predict.

Can you please explain how the above thing actually works? and it would be really helpful if you could supply some additional resources to truly understand the NULL .

Fiddle

Answer 1

This is because NULL s are treated specially in SQL. Whenever you compare something to NULL it returns unknown , which in turn fails the comparison in NOT IN condition.

This is one of the reasons why it is recommended to use EXISTS() over NOT IN .

Fiddle

Answer 2

Do not use NOT IN with subqueries! When the subquery returns a NULL value, it has the behavior you see. No rows are returned.

I strongly recommend NOT EXISTS or LEFT JOIN / WHERE for this reason. So:

select a
from xo
where not exists (select 1 from xo xo2 where xo2.c = xo.a);

This will return the complement of:

select a
from xo
where exists (select 1 from xo xo2 where xo2.c = xo.a);

The technical reason this happens is because NULL is an unknown value, not a special value. So:

1 in (1, 2, 3) returns TRUE
1 not in (1, 2, 3) returns FALSE

But:

1 in (1, 2, 3, NULL) returns TRUE, because 1 is explicitly in the list
1 in (2, 3, NULL) returns NULL (equivalent to FALSE) because `NULL` could be 1 or not

And:

1 not in (1, 2, 3, NULL) returns FALSE because 1 is explicitly in the list
1 not in (2, 3, NULL) returns NULL, because 1 could be in the list

So, when any value is NULL , NOT IN returns either FALSE or NULL -- and hence all rows are filtered out.