Question

select
    day,
    count(*) OVER(PARTITION BY day) as cnt
from a.table_b
where day between '2017-10-13' and '2017-10-14';

此查询的结果如下所示：

day        cnt
2017-10-13  12
2017-10-13  12
2017-10-14  13
2017-10-14  13

但是，当使用group by对此结果进行分组时，

select day, cnt
from 
(
    select day, count(*) OVER(PARTITION BY day) as cnt 
    from a.table_b 
    where day between '2017-10-13' and '2017-10-14'
) t
group by day, cnt;

这是结果：

2017-10-13  1
2017-10-14  1

这个结果显然是错误的，这是一个错误还是错误的用法？

Answer 1

为什么不必要 OVER（PARTITION BY 只使用GROUP BY。

select
    day,
    count(*) as cnt
from a.table_b
where day between '2017-10-13' and '2017-10-14'
gtoup by  day;

hive基本组遍及（分区依据）功能遇到了一个bug

1 个答案: