SQL聚合功能

Question

Given the following table

Month        Product  Sales
---------    -------  -----    
January         1      10
January         2      15
February        3      23
March           1      15

The query I've used is:

SELECT DISTINCT a.* FROM( -- possibly needed for where clause
    SELECT dateadd(month, x.MonthOffset,0) AS mes, x.produto, Sum(x.quantidade) AS vendas
    FROM
    (
        SELECT DATEDIFF(month, 0, a.data) AS MonthOffset, a.produto, a.quantidade
        FROM (
          SELECT l.*, fc.data FROM LINHAS AS l JOIN FacturaCli AS fc
              ON fc.codigo = l.codigo 
              UNION ALL
          SELECT l.*, ff.dataEmissão FROM LINHAS AS l JOIN FacturaForn AS ff
              ON ff.codigo = l.codigo
        ) AS a
    ) AS x
    GROUP BY MonthOffset, produto
) AS a

I would like to get the product with more sales per month.

I'm having trouble doing query that requires a field that must not be in "aggregate group". Note that the table I provided is a result from a query.

I'm using SQL Server 2008. I can't use temporary variables or over clause. I wouldn't like to repeat the query that lead me to this table! But if it is the only way, it's ok.

Expected result:

Month        Product  Sales
---------    -------  -----    
January         2      15
February        3      23
March           1      15

Answer 1

A very tricky solution for your problem.

SELECT a.* 
FROM   mytable a 
       LEFT JOIN mytable b 
              ON a.Month = b.Month
                 AND a.Sales < b.Sales
WHERE  b.Month IS NULL