这是一个快速查询来说明行为:
select
v,
-- FIRST_VALUE() and LAST_VALUE()
first_value(v) over(order by v) f1,
first_value(v) over(order by v rows between unbounded preceding and current row) f2,
first_value(v) over(order by v rows between unbounded preceding and unbounded following) f3,
last_value (v) over(order by v) l1,
last_value (v) over(order by v rows between unbounded preceding and current row) l2,
last_value (v) over(order by v rows between unbounded preceding and unbounded following) l3,
-- For completeness' sake, let's also compare the above with MAX()
max (v) over() m1,
max (v) over(order by v) m2,
max (v) over(order by v rows between unbounded preceding and current row) m3,
max (v) over(order by v rows between unbounded preceding and unbounded following) m4
from (values(1),(2),(3),(4)) t(v)
可以在这里看到上述查询的输出(此处为SQLFiddle):
| V | F1 | F2 | F3 | L1 | L2 | L3 | M1 | M2 | M3 | M4 |
|---|----|----|----|----|----|----|----|----|----|----|
| 1 | 1 | 1 | 1 | 1 | 1 | 4 | 4 | 1 | 1 | 4 |
| 2 | 1 | 1 | 1 | 2 | 2 | 4 | 4 | 2 | 2 | 4 |
| 3 | 1 | 1 | 1 | 3 | 3 | 4 | 4 | 3 | 3 | 4 |
| 4 | 1 | 1 | 1 | 4 | 4 | 4 | 4 | 4 | 4 | 4 |
很少有人会想到应用于带有ORDER BY
子句的窗口函数的隐式框架。在这种情况下,窗口默认为 frame RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW
。(RANGE 与 ROWS 并不完全相同,但那是另一回事了)。这样想:
- 在
v = 1
有序窗口的框架跨度的行上v IN (1)
- 在
v = 2
有序窗口的框架跨度的行上v IN (1, 2)
- 在
v = 3
有序窗口的框架跨度的行上v IN (1, 2, 3)
- 在
v = 4
有序窗口的框架跨度的行上v IN (1, 2, 3, 4)
如果你想阻止这种行为,你有两个选择:
- 对有序窗口函数使用显式
ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING
子句
ORDER BY
在那些允许省略它们的窗口函数中使用 no子句(as MAX(v) OVER()
)
更多细节在这篇文章中解释了关于LEAD()
, LAG()
,FIRST_VALUE()
和LAST_VALUE()