SQL-选择2列的最新不同组合

Question

我有一个带有以下各列的SQL表。 这4行适用于1位用户。

| id | user_id | role_id | preference_order | updated_at          |
|----|---------|---------|------------------|---------------------|
| 83 | 45      | 7       | 1                | 2016-10-24 03:29:18 |
| 84 | 45      | 10      | 2                | 2016-10-24 03:29:18 |
| 85 | 45      | 7       | 1                | 2016-10-25 05:40:18 |
| 86 | 45      | 10      | 2                | 2016-10-25 05:40:18 |

选择user_id和role_id的最新不同组合的快速方法是什么-那么第85行和第86行？

---

我可以得到的最接近的查询是此查询，但它改为获取行83和84：

SELECT * 
FROM user_role UR
GROUP BY UR.user_id, UR.role_id

我也尝试过，它得到85和86，但是对于一张大桌子来说似乎太慢了：

SELECT *
FROM user_role UR
JOIN (SELECT user_id, role_id, max(updated_at) updated_at FROM user_role GROUP BY user_id, role_id) UR2
    ON UR.user_id = UR2.user_id AND UR.role_id = UR2.role_id AND UR.updated_at = UR2.updated_at
ORDER BY UR.order ASC;

Answer 1

如果需要根据最大updated_at的多个user_id，role_id，则可以尝试此查询：

SELECT DISTINCT user_id,role_id,MAX(updated_at) from user_role group by user_id,role_id;

Answer 2

SELECT DISTINCT *
  FROM user_role
       NATURAL JOIN
       ( SELECT user_id, role_id, 
                MAX( updated_at ) AS updated_at 
           FROM user_role
          GROUP
             BY user_id, role_id ) AS t;

Answer 3

利用row_number（）并按user_id和role_id对其进行分区

select  *
from
(
    select  *, rn = row_number () over (partition by user_id, role_id 
                                            order by updated_at desc)
    from    your_table 
) t
where   t.rn    = 1

Answer 4

尝试这个

            select 
            id , user_id , role_id , ord, updated_at
            from
            (
            SELECT * ,
            ROW_NUMBER() OVER ( PARTITION  BY ID ORDER BY  pdated_at DESC ) as RID
            FROM user_role UR
            )A 
            where RID in ('1','2') 

Answer 5

在子查询中按update_at排序，然后在主查询中按分组。

select * from (
  select * from user_role order by user_id asc ,role_id asc,updated_at desc
)UR GROUP BY UR.user_id, UR.role_id

Answer 6

SELECT * 
FROM user_role UR
    LEFT JOIN user_role UR2 ON UR.user_id=UR2.user_id 
                           and UR.role_id=UR2.role_id 
                           and UR.updated_at<UR2.updated_at
WHERE UR2.updated_at IS NULL