假设我有一个SQL查询

SELECT Label1, Label2, SUM(number) as MySum
FROM MyTable
GROUP BY Label1, Label2

它给我一些结果，例如

Label1 | Label2 | MySum 
-------+--------+------
Foo    | Bar    |   100
Foo    | Baz    |  -100
NotFoo | Bar    |   100
NotFoo | Baz    |  -100

我可以放一个WHEREandHAVING子句来过滤结果

SELECT Label1, Label2, SUM(number) as MySum
FROM MyTable
WHERE Label1 = 'Foo'
GROUP BY Label1, Label2
HAVING MySum > 0

这给了我结果

Label1 | Label2 | MySum 
-------+--------+------
Foo    | Bar    |   100

该WHERE和HAVING被AND隐含一起编。

是否有类似的语法

SELECT Label1, Label2, SUM(number) as MySum
FROM MyTable
WHERE Label1 = 'Foo'
GROUP BY Label1, Label2
OR HAVING MySum > 0

给出结果

Label1 | Label2 | MySum 
-------+--------+------
Foo    | Bar    |   100
Foo    | Baz    |  -100
NotFoo | Bar    |   100

？

我可以做

SELECT Label1, Label2, SUM(number) as MySum
FROM MyTable
WHERE Label1 = 'Foo'
GROUP BY Label1, Label2

UNION

SELECT Label1, Label2, SUM(number)
FROM MyTable
GROUP BY Label1, Label2
HAVING MySum > 0

劣势： 这可能是低效率的，因为数据库必须计算Foo | Bar | 100两次该行，发现它是重复的，然后将其丢弃。

我也可以写

SELECT Label1, Label2, SUM(number) as MySum
FROM MyTable
WHERE Label1 = 'Foo'
GROUP BY Label1, Label2

UNION ALL

SELECT Label1, Label2, SUM(number)
FROM MyTable
WHERE Label1 <> 'Foo'
GROUP BY Label1, Label2
HAVING MySum > 0

缺点： 乍看之下，它开始变得不那么容易理解了（可能是最坏的缺点）。

或者

SELECT Label1, Label2, SUM(number) as MySum
FROM MyTable
GROUP BY Label1, Label2
HAVING MySum > 0
OR Label1 = 'Foo'

缺点： 这将使数据库可以做很多额外的工作，将大量的非Foo标签聚集在一起。