求助，高手帮帮忙。

[hww2004]

有如下表：A,
CUS_NO PRD_NO S_DD UP
====== ======= ========== =====
A BAD001 2007-12-01 2.0
A BAD001 2008-01-21 2.2
A BAD001 2008-02-15 2.1
B BAD001 2008-01-11 2.2
B BAD001 2008-02-22 2.3
C BAD001 2008-02-21 2.2
C BAD001 2008-02-23 2.1
A1 BAD002 2008-01-21 2.2
A2 BAD002 2008-01-21 2.3
A2 BAD002 2008-01-26 2.2
... ... ... ...


希望查询的结果如下：
A BAD001 2007-12-01 2.0
A BAD001 2008-01-21 2.2
B BAD001 2008-01-11 2.2
C BAD001 2008-02-21 2.2
A2 BAD002 2008-01-21 2.3

[pt1314917]

---这个可以了啊。。

select * from ta a where  exists
(select 1 from ta where CUS_NO=a.CUS_NO and PRD_NO=a.PRD_NO and S_DD>a.S_DD)

[pt1314917]

create table tA(CUS_NO varchar(8),PRD_NO varchar(10),S_DD datetime,UP decimal(9,2))
go
insert into tA
select 'A','BAD001','2007-12-01',2.0
union select 'A','BAD001','2008-01-21',2.2
union select 'A','BAD001','2008-02-15',2.1
union select 'B','BAD001','2008-01-11',2.2
union select 'B','BAD001','2008-02-22',2.3
union select 'C','BAD001','2008-02-21',2.2
union select 'C','BAD001','2008-02-23',2.1
union select 'A1','BAD002','2008-01-21',2.2
union select 'A2','BAD002','2008-01-21',2.3
union select 'A2','BAD002','2008-01-26',2.3
union select 'A2','BAD002','2008-01-26',2.2
go

select * from ta a where  exists
(select 1 from ta where CUS_NO=a.CUS_NO and PRD_NO=a.PRD_NO and S_DD>a.S_DD)

[fcuandy]

不知道楼主什么想法

[-狙击手-]

create table A(CUS_NO varchar(8),PRD_NO varchar(10),S_DD datetime,UP decimal(9,2))
go
insert into A
select 'A','BAD001','2007-12-01',2.0
union select 'A','BAD001','2008-01-21',2.2
union select 'A','BAD001','2008-02-15',2.1
union select 'B','BAD001','2008-01-11',2.2
union select 'B','BAD001','2008-02-22',2.3
union select 'C','BAD001','2008-02-21',2.2
union select 'C','BAD001','2008-02-23',2.1
union select 'A1','BAD002','2008-01-21',2.2
union select 'A2','BAD002','2008-01-21',2.3
union select 'A2','BAD002','2008-01-26',2.3
union select 'A2','BAD002','2008-01-26',2.2
go
select * from A
select cus_no,prd_no,s_dd,up 
from A 
where not exists(select * 
                 from A b
                 where datepart(yy,a.S_DD)= datepart(yy,b.S_DD) and a.cus_no=b.cus_no and a.prd_no=b.prd_no and a.s_dd > b.s_dd ) 
and cus_no in (select cus_no from A group by cus_no having count(1) > 1)
order by len(cus_no),cus_no



go
drop table A;
go
/*
cus_no   prd_no     s_dd                                                   up          
-------- ---------- ------------------------------------------------------ ----------- 
A        BAD001     2007-12-01 00:00:00.000                                2.00
A        BAD001     2008-01-21 00:00:00.000                                2.20
B        BAD001     2008-01-11 00:00:00.000                                2.20
C        BAD001     2008-02-21 00:00:00.000                                2.20
A2       BAD002     2008-01-21 00:00:00.000                                2.30

（所影响的行数为 5 行）
*/

[hww2004]

happyflagstone 兄：
我用你的语句确实有问题。