SELECT ename  
FROM scott.emp   
START WITH ename = 'KING'   
CONNECT BY PRIOR empno = mgr;   
 
--得到结果为：  
 
KING  
JONES  
SCOTT  
ADAMS  
FORD  
SMITH  
BLAKE  
ALLEN  
WARD  
MARTIN  
TURNER  
JAMES  
 
 
 
而：  
 
 
 
SELECT SYS_CONNECT_BY_PATH(ename, '>') "Path"   
FROM scott.emp   
START WITH ename = 'KING'   
CONNECT BY PRIOR empno = mgr;  
 
 
 
--得到结果为：  
 
 
 
>KING  
>KING>JONES  
>KING>JONES>SCOTT  
>KING>JONES>SCOTT>ADAMS  
>KING>JONES>FORD  
>KING>JONES>FORD>SMITH  
>KING>BLAKE  
>KING>BLAKE>ALLEN  
>KING>BLAKE>WARD  
>KING>BLAKE>MARTIN  
>KING>BLAKE>TURNER  
>KING>BLAKE>JAMES  
>KING>CLARK  
>KING>CLARK>MILLER 
SELECT ename
FROM scott.emp
START WITH ename = 'KING'
CONNECT BY PRIOR empno = mgr;

--得到结果为：

KING
JONES
SCOTT
ADAMS
FORD
SMITH
BLAKE
ALLEN
WARD
MARTIN
TURNER
JAMES

而：

SELECT SYS_CONNECT_BY_PATH(ename, '>') "Path"
FROM scott.emp
START WITH ename = 'KING'
CONNECT BY PRIOR empno = mgr;

--得到结果为：

>KING
>KING>JONES
>KING>JONES>SCOTT
>KING>JONES>SCOTT>ADAMS
>KING>JONES>FORD
>KING>JONES>FORD>SMITH
>KING>BLAKE
>KING>BLAKE>ALLEN
>KING>BLAKE>WARD
>KING>BLAKE>MARTIN
>KING>BLAKE>TURNER
>KING>BLAKE>JAMES
>KING>CLARK
>KING>CLARK>MILLER
 

其实SYS_CONNECT_BY_PATH这个函数是oracle9i才新提出来的！
它一定要和connect by子句合用！
第一个参数是形成树形式的字段，第二个参数是父级和其子级分隔显示用的分隔符！

START WITH 代表你要开始遍历的的节点！

CONNECT BY PRIOR 是标示父子关系的对应！

如下例子：

view plaincopy to clipboardprint?
select max(  
substr(  
sys_connect_by_path(column_name,',')  
,2)  
)  
from (select column_name,rownum rn from user_tab_columns where table_name ='AA_TEST')  
start with rn=1 connect by rn=rownum ; 
select max(
substr(
sys_connect_by_path(column_name,',')
,2)
)
from (select column_name,rownum rn from user_tab_columns where table_name ='AA_TEST')
start with rn=1 connect by rn=rownum ;

是将列用,进行分割成为一行，然后将首个，去掉，只取取最大的那个数据。

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

下面是别人的例子：

1、带层次关系

view plaincopy to clipboardprint?
SQL> create table dept(deptno number,deptname varchar2(20),mgrno number);  
 
Table created.  
 
SQL> insert into dept values(1,'总公司',null);  
 
1 row created.  
 
SQL> insert into dept values(2,'浙江分公司',1);  
 
1 row created.  
 
SQL> insert into dept values(3,'杭州分公司',2);  
 
1 row created.  
 
SQL> commit;  
 
Commit complete.  
 
SQL> select max(substr(sys_connect_by_path(deptname,','),2)) from dept connect by prior deptno=mgrno;  
 
MAX(SUBSTR(SYS_CONNECT_BY_PATH(DEPTNAME,','),2))  
--------------------------------------------------------------------------------  
总公司,浙江分公司,杭州分公司 
SQL> create table dept(deptno number,deptname varchar2(20),mgrno number);

Table created.

SQL> insert into dept values(1,'总公司',null);

1 row created.

SQL> insert into dept values(2,'浙江分公司',1);

1 row created.

SQL> insert into dept values(3,'杭州分公司',2);

1 row created.

SQL> commit;

Commit complete.

SQL> select max(substr(sys_connect_by_path(deptname,','),2)) from dept connect by prior deptno=mgrno;

MAX(SUBSTR(SYS_CONNECT_BY_PATH(DEPTNAME,','),2))
--------------------------------------------------------------------------------
总公司,浙江分公司,杭州分公司

2、行列转换

如把一个表的所有列连成一行，用逗号分隔:

view plaincopy to clipboardprint?
SQL> select max(substr(sys_connect_by_path(column_name,','),2))  
from (select column_name,rownum rn from user_tab_columns where table_name ='DEPT')  
start with rn=1 connect by rn=rownum ;  
 
MAX(SUBSTR(SYS_CONNECT_BY_PATH(COLUMN_NAME,','),2))  
--------------------------------------------------------------------------------  
DEPTNO,DEPTNAME,MGRNO