TABLE(VARRAY) applied to multiple VARRAY instances - possible ???

Question

Question: How can I make a VARRAY column look like a table, if I want the result
being based on multiple rows.
Example: I have got a table with runs of users. For each run I have one row in RESULTS
where the VARRAY v contains one entry per user. I want to join the contents of
the VARRAY to the corresponding USERS rows for all RUNS I find in the RESULTS table.
CREATE TYPE VECTOR AS VARRAY(1000) OF NUMBER
CREATE TABLE RUNS
RUN_ID NUMBER(10) PRIMARY KEY
INSERT INTO RUNS VALUES(1)
INSERT INTO RUNS VALUES(2)
COMMIT
CREATE TABLE USERS
USER_ID NUMBER(10) PRIMARY KEY
,USER_NAME VARCHAR2(4)
INSERT INTO USERS VALUES(1,'Jim')
INSERT INTO USERS VALUES(2,'John')
COMMIT
CREATE TABLE RESULTS
RUN_ID NUMBER(10) PRIMARY KEY
,V VECTOR -- one entry per user here
INSERT INTO RESULTS VALUES(1, VECTOR(3,4))
INSERT INTO RESULTS VALUES(2, VECTOR(5,6))
COMMIT
-- Now I select the results for one run
-- and make VECTOR look like a table using the TABLE function.
SELECT USERS.USER_NAME, RES.RESULT
FROM
(SELECT ROWNUM AS ROW_NUM, COLUMN_VALUE AS RESULT FROM TABLE(SELECT V FROM RESULTS WHERE RUN_ID=1)) RES
, USERS
WHERE RES.ROW_NUM = USERS.USER_ID
USER RESULT
Jim 3
John 4
-- OK, so far.
-- Now generalize that to all runs
SELECT RUN.RUN_ID, USERS.USER_NAME, RES.RESULT
FROM
(SELECT ROWNUM AS ROW_NUM, COLUMN_VALUE AS RESULT FROM TABLE(SELECT V FROM RESULTS WHERE RUN_ID=RUN.RUN_ID)) RES
, USERS
, RUNS
WHERE RES.ROW_NUM = USERS_USER_ID
-- expected:
RUN_ID USER RESULT
1 Jim 3
1 John 4
2 Jim 5
2 John 6
-- got:
(SELECT ROWNUM AS ROW_NUM, COLUMN_VALUE AS RESULT FROM TABLE(SELECT V FROM RESULTS WHERE RUN_ID=RUN
ORA-00904: "RUN"."RUN_ID": invalid identifier
-- Is there a general approach to solve this if the number of runs is variable
-- given the tables layout cannot be changed ????
Thanks for your thoughts
Nils
DROP TABLE RESULTS
DROP TABLE USERS
DROP TABLE RUNS
DROP TYPE VECTOR
/

Answer 1

Agreed but in case of any updation for user_id column
  1  SELECT r.run_id, u.user_name,
  2           r.column_value result
  3    FROM   users u,
  4          (SELECT r.run_id,
  5                  ROW_NUMBER () OVER (
  6                     PARTITION BY r.run_id
  7                     ORDER BY ROWNUM) user_id,
  8                   column_value
  9            FROM   results r, TABLE (r.v)) r
10*   WHERE  r.user_id = u.user_id
11  /
    RUN_ID USER     RESULT
         1 Jim           3
         1 John          4
         2 Jim           5
         2 John          6
SQL>
SQL> SELECT * FROM users;
   USER_ID USER
         1 Jim
         2 John
SQL> UPDATE users SET user_id=user_id+1;
2 rows updated.
SQL> SELECT r.run_id, u.user_name,
  2           r.column_value result
  3    FROM   users u,
  4          (SELECT r.run_id,
  5                  ROW_NUMBER () OVER (
  6                     PARTITION BY r.run_id
  7                     ORDER BY ROWNUM) user_id,
  8                   column_value
  9            FROM   results r, TABLE (r.v)) r
10    WHERE  r.user_id = u.user_id;
    RUN_ID USER     RESULT
         1 Jim           4
         2 Jim           6
SQL> Khurram