Understanding the GROUP BY statement's behaviour

Question

The question is this..

Table is this..

+--------------------------+---------+------+-----+---------+----------------+
| Field                    | Type    | Null | Key | Default | Extra          |
+--------------------------+---------+------+-----+---------+----------------+
| facility_map_id          | int(10) | NO   | PRI | NULL    | auto_increment |
| facility_map_facility_id | int(10) | NO   | MUL | NULL    |                |
| facility_map_listing_id  | int(10) | NO   |     | NULL    |                |
+--------------------------+---------+------+-----+---------+----------------+

Data is this..

+-----------------+--------------------------+-------------------------+
| facility_map_id | facility_map_facility_id | facility_map_listing_id |
+-----------------+--------------------------+-------------------------+
|             248 |                        1 |                      18 |
|             259 |                        1 |                      19 |
|             206 |                        1 |                      20 |
|             244 |                        1 |                      21 |
|             249 |                        2 |                      18 |
|             207 |                        2 |                      20 |
|             208 |                        3 |                      20 |
|             245 |                        3 |                      21 |
|             260 |                        4 |                      19 |
|             261 |                        5 |                      19 |
|             246 |                        6 |                      21 |
|             250 |                        7 |                      18 |
|             247 |                        8 |                      21 |
+-----------------+--------------------------+-------------------------+

I run the this query :

SELECT facility_map_listing_id 
FROM facility_map 
WHERE facility_map_facility_id IN(1, 2) 
GROUP BY facility_map_listing_id 
HAVING count(DISTINCT facility_map_facility_id) >= 2 

and get this..

+-------------------------+
| facility_map_listing_id |
+-------------------------+
|                      18 |
|                      20 |
+-------------------------+
2 rows in set (0.00 sec)

Which is correct! - but can anyone explain, why the GROUP BY needs to be in the statement?

if it Isnt and I run the same query leaving out the GROUP BY I get..

+-------------------------+
| facility_map_listing_id |
+-------------------------+
|                      18 |
+-------------------------+
1 row in set (0.00 sec)

Can any one explain this to me? Thank you!

Answer 1

Without a group by, an aggregate like count works on the set as a whole. So this query returns either zero or one row:

SELECT facility_map_listing_id 
FROM facility_map 
WHERE facility_map_facility_id IN(1, 2)
HAVING count(DISTINCT facility_map_facility_id) >= 2 

It will return one row if the having condition is met, and an empty set otherwise.

Now, with the group by, it evaluates the having condition for each value of facility_map_listing_id. That can return up to as many rows as there are distinct values of facility_map_listing_id.