Oracle SQL-SCD跟踪随时间变化的属性

我遇到了以下错误：

ORA-30483: window  functions are not allowed here
30483. 00000 -  "window  functions are not allowed here"

我正试图建立一个横向视图，以了解随时间变化对BOOK_NO和WALK_NO属性的影响

我认为一个合理的方法是将有效的|排序到日期，然后将书| no | | | walk | no属性按该值解码……但不确定从这里走到哪里

SQL语句如下：

Select METER_PIT_REF 
,MAX(DECODE( RANK() OVER (PARTITION BY METER_PIT_REF ORDER BY EFFECTIVE_TO DESC),1,BOOK_NO||WALK_NO) )
,MAX(DECODE( RANK() OVER (PARTITION BY METER_PIT_REF ORDER BY EFFECTIVE_TO DESC),2,BOOK_NO||WALK_NO) )
,MAX DECODE( RANK() OVER (PARTITION BY METER_PIT_REF ORDER BY EFFECTIVE_TO DESC),3,BOOK_NO||WALK_NO) )
FROM BW_PIT_BOOKWALK_BRIDGE
WHERE BOOK_NO = 'M04'
GROUP BY METER_PIT_REF

非常感谢您的建议

使用子查询：

Select METER_PIT_REF,
       MAX(CASE WHEN r = 1 THEN BOOK_NO||WALK_NO END),
       MAX(CASE WHEN r = 2 THEN BOOK_NO||WALK_NO END),
       MAX(CASE WHEN r = 3 THEN BOOK_NO||WALK_NO END)
FROM (SELECT b.*,
             RANK() OVER (PARTITION BY METER_PIT_REF ORDER BY EFFECTIVE_TO DESC) as r
      FROM BW_PIT_BOOKWALK_BRIDGE
     ) b
WHERE BOOK_NO = 'M04'
GROUP BY METER_PIT_REF;

---

我还将

DECODE（）

替换为

CASE

。这是与ANSI标准相当的标准。

非常好。非常感谢您的快速回复！