研究一条distinct+order by+limit的SQL执行过程，发现limit影响排序结果的彩蛋

假设有如下表：

mysql> select * from tab;
+----+------+
| id | col1 |
+----+------+
|  1 |    2 |
|  2 |    2 |
|  3 |    5 |
|  4 |    3 |
|  5 |    3 |
|  6 |    4 |
+----+------+
6 rows in set (0.01 sec)

mysql> desc tab;
+-------+---------+------+-----+---------+-------+
| Field | Type    | Null | Key | Default | Extra |
+-------+---------+------+-----+---------+-------+
| id    | int(11) | NO   | PRI | NULL    |       |
| col1  | int(11) | YES  |     | NULL    |       |
+-------+---------+------+-----+---------+-------+
2 rows in set (0.00 sec)

那么，下面的SQL是怎样的执行过程呢？
select distinct id from tab order by col1 desc limit 1,2;
1.先执行select distinct id from tab order by col1 desc，在筛选limit 1,2;？等效于select * from (select distinct id from tab order by col1 desc) a limit 1,2；
2.先执行select * from tab order by col1 desc limit 1,2;，再选出distinct id？等效于select distinct id from (select * from tab order by col1 desc limit 1,2) a；
3.先执行select * from tab order by col1 desc，再从结果集中第一行数据进行去重，直到取到3个数值，然后拿后面2个？

我们先从实际数据来验证一下：
首先，先执行一下select distinct id from tab order by col1 desc limit 1,2;，看看实际的返回结果是什么：

mysql> select distinct id from tab order by col1 desc limit 1,2;
+----+
| id |
+----+
|  6 |
|  5 |