$ sqlplus user/pass 

SQL*Plus: Release 9.2.0.6. - Production on Wed Aug  ::  

Copyright (c) , , Oracle Corporation.  All rights reserved. 

 Connected to:

 Oracle9i Enterprise Edition Release 9.2.0.6. - 64bit Production

 With the Partitioning, OLAP and Oracle Data Mining options

 JServer Release 9.2.0.6. - Production 

SQL> select * from wwm2;        --要更新的表 

TOWN                         ID

 -------------------- ----------

 ww'jj                       111

 llll

 dddd

 lllldf

 lllldf

 dsafdf                      

 ljjjjj

 dsafdf                      

TOWN                         ID

 -------------------- ----------

 ljjjjj                      

 SQL> select * from wwm5;            --更新的条件表

TOWN                         ID

 -------------------- ----------

 lllldf

 test                       

SQL> select wwm2.* from wwm2,wwm5 where wwm2.id=wwm5.id

     /

TOWN                         ID

 -------------------- ----------

 ww'jj                       111

 lllldf

 lllldf

 dsafdf                      

 dsafdf                      

 rows selected.

所以,每次需要更新8条数据就是正确的.

相信程序员是通过以下类似的SQL更新的,这是错误的,因为没有加WHERE

 SQL>  update wwm2 set wwm2.town=(select wwm5.town from wwm5 where wwm5.id=wwm2.id)

     /

 rows updated.

SQL> select * from wwm2;

TOWN                         ID

 -------------------- ----------

 lllldf

 lllldf                      

 lllldf

 lllldf

 lllldf

 lllldf                      

 lllldf                      

TOWN                         ID

 -------------------- ----------

 lllldf                      

  rows selected.

可以看到13条记录被更新,符合条件的更新正确,不符合条件的也更新为NULL.以下是正确的方法

方法一:

SQL> update wwm2

     set town=(select town from wwm5 where wwm5.id=wwm2.id)

     where id=(select wwm5.id from wwm5 where wwm5.id=wwm2.id)

     /

 rows updated.

方法二:    与方法一道理相同,这里需要掌握EXIST的相关用法.

SQL> update wwm2

    set town=(select town from wwm5 where wwm5.id=wwm2.id)

    where exists (select  from wwm5 where wwm5.id=wwm2.id)

  rows updated.

方法三:

SQL> update (select a.town atown,a.id aid,b.town btown,b.id bid from wwm2 a,wwm5 b where a.id=b.id)

     set atown=btown

     /

 set atown=btown

     *

 ERROR at line :

 ORA-: cannot modify a column which maps to a non key-preserved table

  * alter table wwm5 add primary key (id)

 SQL> /

Table altered.

    update (select a.town atown,a.id aid,b.town btown,b.id bid from wwm2 a,wwm5 b where a.id=b.id)

   *  set atown=btown

 SQL> /

 rows updated.

这种方法的局限性就是需要PRIMARY 的支持.

方法四:

   1  declare

   2  cursor cur_wwm is select town,id from wwm5;

   3  begin

   4     for my_wwm in cur_wwm loop

   5     update wwm2 set town=my_wwm.town

   6     where id=my_wwm.id;

   7     end loop;

   8* end;

 SQL> /

PL/SQL procedure successfully completed.

SQL> select * from wwm2;

TOWN                         ID

 -------------------- ----------

 lllldf

 lllldf

 llll

 dddd

 lllldf

 lllldf

 lllldf

 lllldf

 ljjjjj

 lllldf                      

TOWN                         ID

 -------------------- ----------

 lllldf

 ljjjjj                      

这个方法是最灵活的了.

方法五:

注意,方法五只能适用于WWM5是WWM2的子集的时候.

      merge into wwm2

      using (select town,id from wwm5) b

      on (wwm2.id=b.id)

      when matched then update set town=b.town

   * when not matched then insert (town,id) values (null,null)

 SQL> /

 rows merged.

SQL> select * from wwm2;

TOWN                         ID

 -------------------- ----------

                                   ---注意这个地方,被插入了一个空值.因为WWM5的ID=9984在WWM2中不能匹配,根本原因是ORACLE9必须有WHEN NOT MATCHED子句,但是ORACLE10可以不许要,也就是ORACLE10可以不写WHEN NOT MATCHED ,就不必插入NULL值了,为解决这个问题,下一步会DELETE WWM5的ID=,这样一来就不会执行WHEN NOT MATCHED

 lllldf

 lllldf

 llll

 dddd

 lllldf

 lllldf

 lllldf

 lllldf

 ljjjjj                      

TOWN                         ID

 -------------------- ----------

 lllldf

 lllldf

 ljjjjj                      

 rows selected.

SQL> delete from wwm5 where id=;

 row deleted.

SQL>     merge into wwm2

 SQL>      using (select town,id from wwm5) b

 SQL>      on (wwm2.id=b.id)

 SQL>      when matched then update set town=b.town

 SQL>   * when not matched then insert (town,id) values (null,null)

 SQL> /

 rows merged.

       以上就是5种关连更新的例子了,希望能给开发人员解惑.

说明：如果select 子句可以返回多行记录，但返回适合where条件的记录只能是唯一的，否则将会报返回单行的select子句返回多行的错误，因为update只能跟据此处的where子句（内层where)进行相应记录的匹配更新，一次只能是一条。