由于内部错误，Redshift 不支持这种类型的相关子查询模式

Question

当我运行第一个查询时，一切都运行良好：

select uid,
(select top 1 city from UserData where UserData.uid = #uids.uid 
group by city)
from #uids;

问题是当我添加

order by count(city)

时，我收到错误：由于内部错误，不支持这种类型的相关子查询模式

select uid,
(select top 1 city from UserData where UserData.uid = #uids.uid 
group by city
order by count(city))
from #uids;

Answer 1

我会在 RedShift 中使用

row_number()

来完成此操作：

select u.uid, ud.city
from uids u left join
     (select uid, city, count(*) as cnt,
             row_number() over (partition by uid order by count(*) desc) as seqnum
      from UserData ud
      group by uid, city
     ) ud
     on ud.uid = u.uid and seqnum = 1;

您也可以在 SQL Server 中尝试此操作。

Answer 2

该错误表明您的子查询不起作用，这很明显，因为您不能在 SQL 的 ORDER 子句中使用函数。您只能在 SELECT 和 HAVING 子句中使用它们。

如果您想按

count(city)

排序，则应将其添加到带有别名的 SELECT 子句中，并按别名排序。所以你的子查询可以是这样的：

(SELECT city, COUNT(*) AS city_count
FROM UserData
WHERE UserData.uid = #uids.uid
GROUP BY city
ORDER BY city_count
LIMIT 1)

Answer 3

Redshift 中不允许在 GROUP BY 列中引用相关子查询的结果。请在此处查看 AWS 文档：https://docs.aws.amazon.com/redshift/latest/dg/r_corlated_subqueries.html