Produce a range of numbers with select from the gaps of ID column

Question

(1)
Suppose I have a table t which has rows like this:
A B
2 4
6 7
I would like to use select to produce rows with numbers ranging between A column and B column inclusive
for each row, like this:
select ... from t ...
2
3
4
6
7
(2)
Suppose I have ID column which has gaps, I would like to get lowest <N> available numbers between the gaps.
I did research, and I can find the range of the gaps, but I cannot have the numbers listed individually, that is why
I ask question (1). But is there a direct way to get the list instead of going through ranges and the list.
For example, I have ID column which has
2
5
6
7
9
2000000
I would like to get a select query that produces
select ...
3
4
8
10
11
I have a way to get a list from 2 to 2000000 and then minus what we have to get all the
gap numbers, but that is not efficient and could runs out of memory.
PS: Before I post to this group, I did research already on "connect by", with recursive queries.

Answer 1

First of all, I would like to thank jeneesh and Frank for helpful and correct reply. However, due to the a bug of this forum,
when I mark jeneesh's reply as correct, I cannot mark Frank's reply as correct anymore. Therefore I would like to formally
state here that I have tested both solutions and they are both correct. jeneesh and Frank use different approach, connect by
and less join ( <=, I do not know what is the proper name for it ).
Secondly I would like to report my small findings: use connect by, you control the level from 1 to the max number, so you
do not need to use order by, and the performance is better (0.37 seconds version 0.92 seconds). And also it performs better
if I use an intermediate view to limit the result sets. One line of code is worth one thousand words, now I am posting two
versions of the code and highlighted the points I mentioned here.
I am sorry that either I do not know how to format the code or I do not have the capability when posting on this forum.
The code listing does not look good, but I hope you get the idea. I have "plain text" and I am not if I can use fixed width font
and preserve space. After I used the spelling checker, it becomes double spaced and I have to manually delete the blank lines.
/* I learned about { code } after the post and edited the original post */
Thanks again to both jeneesh and Frank.
"connect by" version:
SQL> VARIABLE  new_id_cnt NUMBER
SQL> EXEC        :new_id_cnt := 10;
PL/SQL procedure successfully completed.
Elapsed: 00:00:00.00
SQL> WITH my_table AS
  2  (
  3     SELECT num
  4     FROM   prs_elm
  5     WHERE  1=1
  6  )
  7  ,       my_num   AS
  8  (
  9  SELECT num, ct
10  FROM   (
11                    SELECT num AS num, LEAD(num) OVER (ORDER BY num) - num - 1 AS ct
12                    FROM   ( SELECT 0 AS num
13                             FROM   DUAL
14                             UNION ALL
15                             SELECT num
16                             FROM   my_table
17                             UNION ALL
18                             SELECT ( MAX(num) + :new_id_cnt + 1 ) AS num
19                             FROM   my_table
20                    )
21          )
22  WHERE      ct >= 1
23  AND         ROWNUM <= :new_id_cnt
24  )
25  SELECT     num + level AS available_id
26  FROM       my_num
27  WHERE      ROWNUM  <= :new_id_cnt
28  CONNECT BY level <= ct
29  AND        prior num = num
30  AND        prior sys_guid() is not null
31  ;
AVAILABLE_ID
        3219
        3261
        3264
        3269
        3270
        3275
        3281
        3288
        3289
        3290
10 rows selected.
Elapsed: 00:00:00.37"Less join" version:
SQL> VARIABLE  new_id_cnt NUMBER
SQL> EXEC        :new_id_cnt := 10;
PL/SQL procedure successfully completed.
Elapsed: 00:00:00.00
SQL> WITH my_table AS
  2  (
  3     SELECT num
  4     FROM   prs_elm
  5     WHERE  1=1
  6  )
  7  ,       my_num   AS
  8  (
  9                    SELECT num AS num, LEAD(num) OVER (ORDER BY num) - num - 1 AS ct
10                    FROM   ( SELECT 0 AS num
11                             FROM   DUAL
12                             UNION ALL
13                             SELECT num
14                             FROM   my_table
15                             UNION ALL
16                             SELECT ( MAX(num) + :new_id_cnt + 1 ) AS num
17                             FROM   my_table
18                    )
19  )
20  ,       my_cnt    AS
21  (
22          SELECT  LEVEL AS ct
23          FROM    DUAL
24          CONNECT BY    LEVEL <= :new_id_cnt
25  )
26  SELECT     available_id
27  FROM
28  (
29          SELECT    n.num + c.ct AS available_id
30          FROM      my_num n
31          JOIN      my_cnt c  ON c.ct <= n.ct
32          WHERE     n.ct >= 1
33          ORDER BY  available_id
34  )
35  WHERE      ROWNUM  <= :new_id_cnt
36  ;
AVAILABLE_ID
        3219
        3261
        3264
        3269
        3270
        3275
        3281
        3288
        3289
        3290
10 rows selected.
Elapsed: 00:00:00.92PS: In Frank's code, there is a typo, <= should read as <.
Edited by: PuraVidaOTN on 04-feb-2013 22:49
What: To use tags to format the code.
Edited by: PuraVidaOTN on 04-feb-2013 22:56