Question

我在PostgreSQL 9.4中遇到组内错误：

错误：有序集合排名需要WITHIN GROUP 第4行:(选择a1.rank为r1，

create view a as (select rank() over (order by pid, time)
    pid, event, time
    from test5);



select e1.prev, e1.event, count(*) 
from 
    eventtransitions as e1,
    (select a1.rank as r1,
         a2.rank as r2,
         a1.event as a1_event,
         a2.event as a2_event,
         a1.pid as a1_pid,
         a2.pid as a2_pid
from a as a1, a as a2) as temp
where r1 = r2-1
and e1.event = a2_event
and e1.prev = a1_event
and a1_pid = a2_pid
group by e1.prev, e1.event;

这对我来说有点奇怪，因为根据我对这个错误的理解，这意味着我需要为我的聚合函数等级设置order by子句。但是，我确实将视图设为a，并且我在那里有了order by子句。如何修复此错误以及我做出的假设是错误的？

Answer 1

我认为您的错误在视图中是错误的。我相信你想要的是这个：

create view a as
select
  rank() over (order by pid, time),
  pid, event, time
from test5;

因为它是您当前的视图定义：

create view a as (select rank() over (order by pid, time)
    pid, event, time
    from test5);

实际上接受rank()函数并将其别名为“pid”，这相当于：

create view a as (select rank() over (order by pid, time) as pid,
    event, time
    from test5);

因此，当您在选择查询中调用rank时，它指的是函数，而不是字段（不存在，因为您将其别名化）。

在Postgresql中的组错误中

1 个答案: