Plsql pagination combined with group by.

Question

Hi all,
I am trying to achive something very unusual in a PL/SQL procedure.Consider the following scenario:
I have a table called rhit_rowtest like this:
NAME                            Null?     Type
ENRTASK_ID                                VARCHAR2(30)
REGNO                                     VARCHAR2(30)It has been populated with the following values:(in that order):
insert into rhit_rowtest (enrtask_id,regno) values('5','7');
insert into rhit_rowtest (enrtask_id,regno) values('5','5');
insert into rhit_rowtest (enrtask_id,regno) values('4','6');
insert into rhit_rowtest (enrtask_id,regno) values('2','4');
insert into rhit_rowtest (enrtask_id,regno) values('2','6');
insert into rhit_rowtest (enrtask_id,regno) values('3','7');
insert into rhit_rowtest (enrtask_id,regno) values('3','8');
insert into rhit_rowtest (enrtask_id,regno) values('1','1');
insert into rhit_rowtest (enrtask_id,regno) values('1','2');
insert into rhit_rowtest (enrtask_id,regno) values('1','3');
insert into rhit_rowtest (enrtask_id,regno) values('3','8');
insert into rhit_rowtest (enrtask_id,regno) values('4','9');
insert into rhit_rowtest (enrtask_id,regno) values('4','4');
insert into rhit_rowtest (enrtask_id,regno) values('5','8');
insert into rhit_rowtest (enrtask_id,regno) values('5','10');
insert into rhit_rowtest (enrtask_id,regno) values('2','8');As we can see, there 5 distinct values of enrtask_id (1,2,3,4,5). Now lets say I want to return rows 2 to 4 ordered by enrtask_id and regno.That is pretty simple like this:
select * from (
   select p.*, rownum rnum
           from (select * from rhit_rowtest order by enrtask_id,regno) p
          where rownum <= 4
) where rnum >= 2I get the following values:
          ENRTASK_ID                        REGNO                              RNUM                 
1                              2                                                                   2
1                              3                                                                   3
2                              4                                                                   4 But what I really want to return when 2 and 4 is passed is all the regno values which belong to enrtask_id (2,3,4) like this :
          ENRTASK_ID                        REGNO            
2                                  4
2                                  6
2                                  8
3                                  7
3                                  8
3                                  8
4                                  4
4                                  6
4                                  9So, basically enrtask_id values in this case should each be should be viausalized as a single row and treated as one entity. I believe group by needs to be used somwhow but cant quite figure out how. Any help would be appreciated.
Thanks.

Answer 1

Hi,
Isn't that just
SELECT     *
FROM     rhit_rowtest
WHERE     enrtask_id     BETWEEN     '2'     -- any low value will do
               AND     '4'     -- any higher value
Do you want the 2nd through 4th distinct values (in normal sort order), not necessarily '2' through '4'? That is, if we add this to the sample data:
insert into rhit_rowtest (enrtask_id,regno) values('10','11');Now, since enrtask_id is a VARCHAR2, the value '10' comes after '1' and before '2', so the 2nd, 3rd and 4th values are now '10', '2' and '3' respectively.
Would you now want these results?
ENRTASK_ID REGNO
10         11
2          8
2          4
2          6
3          7
3          8
3          8? If so, you can use the analytic DENSE_RANK fucntion, like this:
WITH     got_r_num     AS
     SELECT     r.*
     ,     DENSE_RANK () OVER (ORDER BY  enrtask_id)     AS r_num
     FROM     rhit_rowtest  r
SELECT     enrtask_id, regno
FROM     got_r_num
WHERE     r_num     BETWEEN     2     -- any low value will do
          AND     4     -- any higher value
;