当我们定义了一个很复杂的cursor,这个cursor需要执行很长时间,在这个cursor执行的过程中,其它用户又修改了该cursor所引用的表的数据时,cursor得到的是修改前还是修改后的数据呢?
答案是cursor得到的始终是cursor在open时的数据,接下来我们将通过一个小实验来验证。
首先,session1执行以下匿名块,该匿名块通过cursor取得t1表的所有数据,不过在open cursor后将暂停30秒,在这30秒中我们将在session2中删除t1表的所有数据:
DECLARE
CURSOR c IS
SELECT deptno
,dname
,loc
FROM t1;
TYPE dept_tab IS TABLE OF c%ROWTYPE;
l_depts dept_tab;
BEGIN
dbms_output.put_line('opening c: ' ||
to_char(SYSDATE, 'yyyy-mm-dd hh24:mi:ss'));
OPEN c;
dbms_lock.sleep(30);
dbms_output.put_line('after sleep: ' ||
to_char(SYSDATE, 'yyyy-mm-dd hh24:mi:ss'));
FETCH c BULK COLLECT
INTO l_depts;
CLOSE c;
FOR i IN l_depts.FIRST .. l_depts.LAST
LOOP
dbms_output.put_line(l_depts(i).deptno || ', ' || l_depts(i)
.dname || ', ' || l_depts(i).loc);
END LOOP;
END;
第二步,session2执行以下语句:
22:35:21 SQL> select * from t1;
DEPTNO DNAME LOC
------ -------------- -------------
10 ACCOUNTING NEW YORK
20 RESEARCH DALLAS
30 SALES CHICAGO
40 OPERATIONS BOSTON
22:35:29 SQL> delete from t1;
4 rows deleted
22:35:33 SQL> commit;
Commit complete
22:35:35 SQL> select * from t1;
DEPTNO DNAME LOC
------ -------------- -------------
22:35:38 SQL>
最后,观察session1的输出:
opening c: 2011-10-26 22:35:25
after sleep: 2011-10-26 22:35:55
10, ACCOUNTING, NEW YORK
20, RESEARCH, DALLAS
30, SALES, CHICAGO
40, OPERATIONS, BOSTON
由于在22:35:25我们就已经打开了游标,所以结果依然能输出t1表的所有数据,尽管在22:35:35之前我们已经删除了t1表的所有数据并提交,而cursor取数据(fetch)发生在22:35:55之后。
ref:
The OPEN statement executes the query associated with a cursor. It allocates database resources to process the query and identifies the result set -- the rows that match the query conditions.
http://download.oracle.com/docs/cd/B19306_01/appdev.102/b14261/open_statement.htm#i35173