SQL & PL/SQL

Imagine that I have an object definition A

CREATE TYPE A
  AS OBJECT (
    col1 NUMBER
);

and a collection tbl_a

CREATE TYPE tbl_a AS TABLE OF a;

Now, assume that I need to have two pipelined table functions (F1 and F2) that return TBL_A collections. These functions will share the same logic, so I want one function to call the other. Assuming that F1 is defined as

CREATE FUNCTION f1( arg IN NUMBER )
  RETURN tbl_a
  PIPELINED
IS
BEGIN
  FOR x IN 1..100
  LOOP
    PIPE ROW( new a( x ) );
  END LOOP;
  RETURN;
END;

then I can define F2 as

CREATE FUNCTION f2( arg IN NUMBER )
  RETURN tbl_a
  PIPELINED
IS
  local_a a;
BEGIN
  FOR x IN (SELECT * FROM TABLE( f1( arg ) ) )
  LOOP
    local_a := new a( x.col1 );
    PIPE ROW( local_a );
  END LOOP;
  RETURN;
END;

Is there a way to accomplish the same thing but without having to go through the effort of creating a new instance of my object in F2? Essentially, if the object definition changes, I would like to be able to just change F1 without worrying about F2.

One option would be to create a helper function that returns a non-pipelined collection of objects and have cookie-cutter logic in F1 & F2 to pipeline the collection, i.e.

CREATE FUNCTION helper_func( arg IN NUMBER )
  RETURN tbl_a
IS
...

CREATE FUNCTION f1( arg IN NUMBER )
  RETURN tbl_a
  PIPELINED
IS
  local_table_a tbl_a := helper_func( arg );
BEGIN
  FOR x IN local_table_a.FIRST .. local_table_a.LAST
  LOOP
    PIPE ROW( local_table_a(x) );
  END LOOP;
  RETURN;
END;

This, however, requires that we read the entire collection into memory at once, which would use a bit more RAM than we're comfortable using.

Any other options?

Thanks,

Justin