oracle merge从9i到10g的增强

    MERGE语句是Oracle9i新增的语法，用来合并UPDATE和INSERT语句。连接条件匹配上的进行UPDATE，无法匹配的执行INSERT。这个语法仅需要一次全表扫描就完成了全部工作，执行效率要高于INSERT＋UPDATE。



    下面看个具体例子：

hr@ORCL> select * from p;

ID NAME
---------- ----------
1 d
3 e
8 f

hr@ORCL> select * from t;

ID NAME
---------- ----------
1 a
2 b
3 c

hr@ORCL> merge into t using p
2         on (t.id=p.id)
3       when matched then
4     update set
5       t.name=p.name
6       when not matched then
7     insert values
8       (p.id,p.name)
9  ;

3 rows merged.

hr@ORCL> select * from t;

ID NAME
---------- ----------
1 d
2 b
3 e
8 f

hr@ORCL> commit;

Commit complete.

    下面稍微修改一下例子：

hr@ORCL> insert into p values(1,'h');

1 row created.

hr@ORCL> commit;

Commit complete.

hr@ORCL> merge into t using p
2         on (t.id=p.id)
3       when matched then
4     update set
5       t.name=p.name
6       when not matched then
7     insert values
8       (p.id,p.name)
9  ;
merge into t using p
*
ERROR at line 1:
ORA-30926: unable to get a stable set of rows in the source tables

    这个错误是使用MERGE最常见的错误，造成这个错误的原因是由于通过连接条件得到的源表的记录不唯一。在merge into时需要设定一个key值，会根据这个key值来决定merge into的操作（update还是insert into），所以要求在merge时这个key值是唯一的。所以要先从源表中选出全部数据而且key值是唯一的。

    还有一个错误也是比较常见的：

hr@ORCL> create table mm (id number,name varchar2(10));

Table created.

hr@ORCL> create table mn (id number,name varchar2(10));

Table created.

hr@ORCL> insert into mm values(1,'A');

1 row created.

hr@ORCL> insert into mn values(1,'B');

1 row created.

hr@ORCL> merge into mn using mm
2         on (mn.id=mm.id)
3       when matched then
4     update set mn.id=mm.id
5       when not matched then
6     insert values
7      (mm.id,mm.name);
on (mn.id=mm.id)
*
ERROR at line 2:
ORA-38104: Columns referenced in the ON Clause cannot be updated: "MN"."ID"

    10g增强一：where子句

    下面看一个小例子：

hr@ORCL> select * from p;

ID NAME
---------- ----------
1 d
2 f
7 g

hr@ORCL> select * from t;

ID NAME
---------- ----------
1 a
2 b
3 c

hr@ORCL> merge into t using p
2         on (p.id=t.id)
3      when matched then
4        update set
5          t.name=p.name
6        where p.id=2;

1 row merged.

hr@ORCL> select * from t;

ID NAME
---------- ----------
1 a
2 f
3 c

    10g增强二：delete子句

hr@ORCL> select * from p;

ID NAME
---------- ----------
1 d
2 f
7 g

hr@ORCL> select * from t;

ID NAME
---------- ----------
1 a
2 f
3 c

hr@ORCL> merge into t using p
2         on (p.id=t.id)
3       when matched then
4     update set
5         t.name=p.name
6     delete where
7        (t.id=2);

2 rows merged.

hr@ORCL> select * from t;

ID NAME
---------- ----------
1 d
3 c