合并sql，请大家帮帮啊，谢谢

[心如刀割]

select a.id,a.dt,a.val,b.sumVal from ( 
select id,dt, val,TO_CHAR(TO_DATE(dt), 'YYYY/MM')   as mon from test a 
)a ,( 
select id,sum(val) as sumVal,TO_CHAR(TO_DATE(dt), 'YYYY/MM')  as mon from test  group by TO_CHAR(TO_DATE(dt), 'YYYY/MM')  ,id  
)b 
where a.id= b.id and a.mon = b.mon 
order by a.id,a.dt; 

这个sql就是根据id查出每天的val和每个月sum出来的val。不知道能不能换种写法，因为查每天的val去检索了一次test表，查每个月sum的val又去检索了test 表。如果数据量大的话很消耗性能，这还是没有加条件的。如果在每个test表后面加上大量的条件，非常慢。能不能合并成一个？？？

[小灰狼W]

分析函数
sum(val)over(partition by id,dt)

select t.id,t.dt,sum(t.val)over(partition by t.id,t.dt) sumVal
from test t
order by ...

[小灰狼W]

实际上，条件够苛刻的话性能会更好...没有条件才是最糟糕的

[304的的哥]

--分析函数，sum(val) over (partition by to_char(dt,'yyyy-mm'))
SQL> with t as(
  2  select 1 id,to_date('2011-01-01','yyyy-mm-dd') dt,50 val from dual union all
  3  select 2,to_date('2011-01-05','yyyy-mm-dd'),20 from dual union all
  4  select 3,to_date('2011-02-05','yyyy-mm-dd'),500 from dual union all
  5  select 5,to_date('2011-02-09','yyyy-mm-dd'),100 from dual union all
  6  select 6,to_date('2011-02-25','yyyy-mm-dd'),5050 from dual union all
  7  select 7,to_date('2011-03-20','yyyy-mm-dd'),10000 from dual)
  8  select id,dt,val,sum(val) over (partition by to_char(dt,'yyyy-mm')) sumval
  9  from t
 10  group by id,dt,val
 11  /
 
        ID DT                 VAL     SUMVAL
---------- ----------- ---------- ----------
         1 2011-01-01          50         70
         2 2011-01-05          20         70
         3 2011-02-05         500       5650
         5 2011-02-09         100       5650
         6 2011-02-25        5050       5650
         7 2011-03-20       10000      10000
 
6 rows selected

[心如刀割]

太感谢了。。