cursor loop and continue statement : unexpected behaviour

From: rob <"spamlovelyspamwonderfulspam>
Date: Tue, 09 Jun 2009 10:21:24 +0200
Message-ID: <9f29a$4a2e1b84$82a1e2fc$1492_at_news1.tudelft.nl>



Hi all,

I might be overlooking something due to deadline stress. But this behaviour amazes me. It looks as if the cursor caches 100 rows and the continue statement flushes the cache and begins with the first record of a new cache fetch.

I narrowed it down to the following script:

drop table test1;

create table test1 (test1_id NUMBER);

begin

   for i in 1..300
   loop
     insert into test1 values (i);
   end loop;
end;
/

declare

   cursor c_test1 is
   select *
   from test1;
begin

   for c in c_test1
   loop

      if mod(c.test1_id,10) = 0
      then
       dbms_output.put_line(c_test1%ROWCOUNT||' '||c.test1_id||' Continue');
       continue;
      end if;
      dbms_output.put_line(c_test1%ROWCOUNT||' '||c.test1_id||' Process');
   end loop;
end;
/

1 1 Process
2 2 Process
3 3 Process
4 4 Process
5 5 Process
6 6 Process
7 7 Process
8 8 Process
9 9 Process
10 10 Continue << Where are tes1_id's 11 to 100? >> 11 101 Process
12 102 Process
13 103 Process
14 104 Process
15 105 Process
16 106 Process
17 107 Process
18 108 Process
19 109 Process
20 110 Continue << Where are tes1_id's 111 to 200? >> 21 201 Process
22 202 Process
23 203 Process
24 204 Process
25 205 Process
26 206 Process
27 207 Process
28 208 Process
29 209 Process
30 210 Continue << Where are tes1_id's 211 to 300? >>

Oracle Database 11g Enterprise Edition Release 11.1.0.7.0 - Production PL/SQL Release 11.1.0.7.0 - Production
redhat release 5
2 node RAC Received on Tue Jun 09 2009 - 03:21:24 CDT

Original text of this message