LEFT JOIN 如何比未加入的查询返回更多的结果？

Question

考虑以下查询：

mysql> SELECT count(*) FROM list l WHERE l.source='blink';
+----------+
| count(*) |
+----------+
|     3372 |
+----------+
1 row in set (0.00 sec)

mysql> SELECT count(*) FROM list l LEFT JOIN cardinal c ON l.id=c.id WHERE c.resolution IN ('Left','Right','Up') AND l.source='blink';
+----------+
| count(*) |
+----------+
|     5116 |
+----------+
1 row in set (2.47 sec)

INSERT在这两个运行的查询之间没有在数据库上运行或运行其他查询。第二个查询似乎比第一个查询更严格，那么它怎么会返回更多的行呢？请注意，该list表有超过 200 万行，而该cardinal表大约有 50 万行。

编辑：添加EXPLAIN输出：

mysql> EXPLAIN SELECT count(*) FROM list l LEFT JOIN cardinal c ON l.id=c.id WHERE c.resolution IN ('Left','Right','Up') AND l.source='blink';
+----+-------------+-------+--------+-------------------+---------+---------+-------------+--------+-------------+
| id | select_type | table | type   | possible_keys     | key     | key_len | ref         | rows   | Extra       |
+----+-------------+-------+--------+-------------------+---------+---------+-------------+--------+-------------+
|  1 | SIMPLE      | c     | ALL    | NULL              | NULL    | NULL    | NULL        | 536258 | Using where |
|  1 | SIMPLE      | l     | eq_ref | PRIMARY,id,source | PRIMARY | 8       | direct.c.id |      1 | Using where |
+----+-------------+-------+--------+-------------------+---------+---------+-------------+--------+-------------+
2 rows in set (0.00 sec)

mysql> select count(*) from list;
+----------+
| count(*) |
+----------+
|  2165664 |
+----------+
1 row in set (1.32 sec)

mysql> select count(*) from cardinal;
+----------+
| count(*) |
+----------+
|   537007 |
+----------+
1 row in set (0.23 sec)

Answer 1

第一个表中的每一行可能与第二个表中的多行匹配。

您可以通过以下方式检查：

SELECT l.id,count(*) 
FROM list l 
LEFT JOIN cardinal c ON l.id=c.id 
WHERE c.resolution IN ('Left','Right','Up') AND l.source='blink' 
group by l.id 
having count(*)>1;

一个小的演示是在SQL提琴这里。