使用over子句运行总计

Question

我正在运行SQL Server 2008,我目前正在尝试更多地理解over子句,我之前使用过它,但row_number()我知道你可以使用聚合函数并创建像运行总计这样的东西.

我创建了以下测试数据库.

CREATE TABLE sales
(
    employee nvarchar(50),
    sales decimal(18, 2),
    datesale datetime
);

Insert into sales Values('John','54.23','2017-01-30 08:00:00.000')
Insert into sales Values('John','123.96','2017-01-30 09:00:00.000')
Insert into sales Values('John','534.20','2017-01-30 10:00:00.000')
Insert into sales Values('Ben','98.34','2017-01-29 04:00:00.000')
Insert into sales Values('Ben','126.32','2017-01-29 05:00:00.000')

并使用以下查询为每个员工创建一个运行总和

select 
    employee, sales, datesale, 
    SUM(sales) over (partition by employee order by datesale) as mvgsum 
from sales

在使用SQL Server 2008的SQL小提琴上,我得到以下结果

 | employee |  sales |                  datesale | mvgsum |
 |----------|--------|---------------------------|--------|
 |      Ben |  98.34 | January, 29 2017 04:00:00 |  98.34 |
 |      Ben | 126.32 | January, 29 2017 05:00:00 | 224.66 |
 |     John |  54.23 | January, 30 2017 08:00:00 |  54.23 |
 |     John | 123.96 | January, 30 2017 09:00:00 | 178.19 |
 |     John |  534.2 | January, 30 2017 10:00:00 | 712.39 |

但是,在SQL Server Management Studio中,在SQL Server 2008数据库上,我使用相同的查询获得以下错误,具有相同的表结构和数据.

'order'附近的语法不正确.

可能是什么导致了这个？

Answer 1

SUM() OVER() SQL Server 2008不支持.

您可以使用相关子查询:

select employee,
    sales,
    datesale,
    (
        select sum(sales)
        from sales s2
        where s1.employee = s2.employee
            and s2.datesale <= s1.datesale
        ) mvgsum
from sales s1

或者CROSS APPLY:

select employee,
    sales,
    datesale,
    x.mvgsum
from sales s1
cross apply (
    select sum(sales) mvgsum
    from sales s2
    where s1.employee = s2.employee
        and s2.datesale <= s1.datesale
    ) x