5

使用 Oracle 11g 第 2 版，以下查询给出 ORA-01790：表达式必须具有与相应表达式相同的数据类型：

with intervals(time_interval) AS
 (select trunc(systimestamp)
    from dual
  union all
  select (time_interval + numtodsinterval(10, 'Minute'))
    from intervals
   where time_interval < systimestamp)
select time_interval from intervals;

该错误表明 UNION ALL 的两个子查询的数据类型都返回不同的数据类型。

即使我在每个子查询中强制转换为 TIMESTAMP，也会得到相同的错误。

我错过了什么？

编辑：我不是在寻找 CONNECT BY 替代品。

4

3 回答 3

8

在我看来，“递归子查询分解”在 11g R2 中被破坏，用于带有日期或时间戳列的查询。

with test(X) as
(
  select to_date('2010-01-01','YYYY-MM-DD') from dual
  union all (
    select (X + 1) from test where X <= to_date('2010-01-10','YYYY-MM-DD') 
  )
)
select * from test;

ORA-01790

使用强制转换来转换数据类型：

with test(X) as
(
  select cast(to_date('2010-01-01','YYYY-MM-DD') as date) from dual
  union all (
    select (X + 1) from test where X <= to_date('2010-01-10','YYYY-MM-DD') 
  )
)
select * from test;

X
-------------------
2010-01-01 00:00:00

1 row selected

将日期转换为日期会有所帮助，但其他结果在哪里？

它变得更好......

尝试使用另一个开始日期：

with test(X) as
(
  select cast(to_date('2007-01-01','YYYY-MM-DD') as DATE) from dual
  union all (
    select (X + 1) from test where X <= to_date('2011-01-11','YYYY-MM-DD') 
  )
)
select * from test 
where rownum < 10; -- important!

X
-------------------
2007-01-01 00:00:00
2006-12-31 00:00:00
2006-12-30 00:00:00
2006-12-29 00:00:00
2006-12-28 00:00:00
2006-12-27 00:00:00
2006-12-26 00:00:00
2006-12-25 00:00:00
2006-12-24 00:00:00

9 rows selected

倒数？为什么？

2014 年 1 月 14 日更新：作为一种解决方法，使用从结束日期开始的 CTE 并向后构建递归 CTE，如下所示：

with test(X) as
(
  select cast(to_date('2011-01-20','YYYY-MM-DD') as DATE) as x from dual
  union all (
    select cast(X - 1 AS DATE) from test 
    where X > to_date('2011-01-01','YYYY-MM-DD') 
  )
)
select * from test

结果：

|                              X |
|--------------------------------|
| January, 20 2011 00:00:00+0000 |
| January, 19 2011 00:00:00+0000 |
| January, 18 2011 00:00:00+0000 |
| January, 17 2011 00:00:00+0000 |
| January, 16 2011 00:00:00+0000 |
| January, 15 2011 00:00:00+0000 |
| January, 14 2011 00:00:00+0000 |
| January, 13 2011 00:00:00+0000 |
| January, 12 2011 00:00:00+0000 |
| January, 11 2011 00:00:00+0000 |
| January, 10 2011 00:00:00+0000 |
| January, 09 2011 00:00:00+0000 |
| January, 08 2011 00:00:00+0000 |
| January, 07 2011 00:00:00+0000 |
| January, 06 2011 00:00:00+0000 |
| January, 05 2011 00:00:00+0000 |
| January, 04 2011 00:00:00+0000 |
| January, 03 2011 00:00:00+0000 |
| January, 02 2011 00:00:00+0000 |
| January, 01 2011 00:00:00+0000 |

进行的测试：

Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - 64bit Production
于 2010-05-17T17:36:54.570 回答
2

奇数 - 如果您传递varchars 并转换（不强制转换），则有效：

WITH intervals(time_interval) AS
  (SELECT to_char(TRUNC(systimestamp))
  FROM dual
  UNION ALL
  SELECT to_char(to_timestamp(time_interval) + numtodsinterval(10, 'Minute'))
  FROM intervals
  WHERE to_timestamp(time_interval) < systimestamp
  )
SELECT to_timestamp(time_interval) time_interval
FROM intervals
于 2011-09-07T21:52:00.113 回答
0

我不知道类型不匹配，但这是完成我认为你想要的另一种方法（在 10gr2 中有效）：

select base_time + numtodsinterval( 10*(level-1), 'Minute')
from (select trunc(systimestamp) base_time from dual)
connect by base_time + numtodsinterval( 10*(level-1), 'Minute') < systimestamp
于 2010-04-06T19:50:18.820 回答