我有一个视图，其中结果按序列的降序返回。结果集中的一列是correlation_id，对于一堆行是相同的。我希望按照创建时间戳的顺序在correlation_id的分区中应用排序。
Data:
---------------------------
SEQ | CORRELATION_ID | CR_TIMESTAMP
9 | Z | 22/FEB/16 03:00:19.191000000 PM
8 | Z | 22/FEB/16 02:00:26.577000000 PM
7 | Z | 22/FEB/16 01:07:58.171000000 PM
6 | A | 22/FEB/16 03:07:58.171000000 PM
5 | A | 22/FEB/16 02:07:58.171000000 PM
What I want is maintain original order, only sort within the partition by CR_TIMESTAMP:
---------------------------
SEQ | CORRELATION_ID | CR_TIMESTAMP | SRLNO
7 | Z | 22/FEB/16 03:07:58.171000000 PM | 1
8 | Z | 22/FEB/16 02:00:26.577000000 PM | 2
9 | Z | 22/FEB/16 01:07:58.171000000 PM | 3
5 | A | 22/FEB/16 02:07:58.171000000 PM | 1
6 | A | 22/FEB/16 03:07:58.171000000 PM | 2
I tried (without success)
select V.*, ROW_NUMBER()
OVER (PARTITION BY CORRELATION_ID
ORDER BY CR_TIMESTAMP ASC) as SRLNO FROM A_VIEW V;
我的尝试结果为：最终结果按CORRELATION_ID的升序排列。即所有A，然后是所有B，然后是......所有Z.在每个分区中，排名正确地按照CR_TIMESTAMP的顺序排列。
答案 0 :(得分：1)
以下是我理解您的请求的方法：您首先需要CORRELATION_ID Z，因为其最高SEQ（9）高于A的最高SEQ（6），但在每个CORRELATION_ID内{ {1}}您希望按日期排序记录。
select seq, correlation_id, cr_timestamp
from mytable
order by max(seq) over (partition by correlation_id) desc, cr_timestamp desc;