Well Dave, its like this ..
I am trying to write a module which can give the
functionality of a database created using arrays and hashes ..
i.e. create a virtual in memory database using arrays and hashes which could act as tables.
And then allow running simple SQL statements to process data in these arrays/hashes just like database tables.

eg. 1
If the user choses to create the database using one "two dimensional array" named sales,
I want him to be able to process the data in array sales using a simple query like ..
select 0,sum(1) from sales group by 0 order by 0
where 0 n 1 are the columns of array sales.
O/P would be an array.

This avoids the need for writing nested for loops in the calling program ..

eg. 2
user creates a database using a one "two dimensional" salary array and one "hash" parameters..
Then he could use a query like below to select all the rows
from array salary which have the salary value above a specified min salary.
select salary.* from salary,parameters where salary.1 > parameters.minsal order by salary.0

Ofcourse these are hypothetical examples and one would be
better off getting processed o/p from the database itself.

But this module which I am writing is for cases where my datasource is not database,
and I need to have the ease of processing my datastructures like database tables.
Let me know if more details are required.