Select data from plsql collections

Question

Hi All,
I am not a developer but working as a DBA, so not very much familiar with pl/sql, still gone through with documentation and could come up with some solution of my problem. I need some expert advice here.
Problem : I am writing down some kind of plsql program for monitoring of some special batch job, I know we have lot of other option to do the same including db/grid control ..etc but for some
reason i have to do this using plsql only.
Requirement : my requirement is to select data from table in plsql and then should have ability to query it again and again. I would not prefer to go to table rather than directly from plsql..
I wrote down below code for sample, bulk collect data into collection type and can print using for loop.
Declare
type ts is table of v$session%rowtype index by pls_integer;
tsess ts;
begin
select * bulk collect into tsess from v$session ;
for i in 1..tsess.count loop
dbms_output.put_line(tsess(i).terminal);
end loop;
end;
But, is there any way same collection ( tsess in above example ) can be queried using select statement like 'select * from table ( Tsess ) ' I have searched on net and found this can be done using creating type at database level. But my problem is I can not create any object in database as being it is a production one.
I was looking for if is there any way same can be accomplished ... like cast / multiset .. however, I could not get it through.
your help would be appreciated !!
Regards,

Answer 1

I don't think you need to use arrays here, only SQL, have a look at subquery factors and report back if that is insufficient...
Edited by: BrendanP on 12-Feb-2012 03:07 to add an example:
I gather that you want to 'requery' data that you have already got from the database purely to be able to use SQL functionality such as ORDER BY in multiple ways. Here is how you can do this in the original SQL for one particular example, where you want to query v$sql ordering by CPU time and by disk reads separately (I tested this but the output won't look good here, so omitting it):
WITH v AS (
SELECT
    Substr (sql_text,1,500)             sql_text,
    cpu_time/1000000                    cpu_seconds,
    disk_reads,
    buffer_gets,
    executions,
    CASE WHEN rows_processed != 0 THEN Round( buffer_gets / Nvl (Replace (rows_processed, 0, 1) ,1)) END Buffer_gets_rows_proc,
    Round (buffer_gets / Nvl (Replace (executions, 0, 1), 1)) Buffer_gets_executions,
    elapsed_time / 1000000              elapsed_second,
    module
FROM v$sql s)
SELECT
    'CPU'                order_by,
    cpu_seconds          order_val,
    sql_text,
    cpu_seconds,
    disk_reads,
    buffer_gets,
    executions,
    buffer_gets_rows_proc,
    buffer_gets_executions,
    elapsed_second,
    module
FROM v
UNION
SELECT
    'Disk reads',
    disk_reads,
    sql_text,
    cpu_seconds,
    disk_reads,
    buffer_gets,
    executions,
    buffer_gets_rows_proc,
    buffer_gets_executions,
    elapsed_second,
    module
FROM v
ORDER BY order_by, order_val DESC