Criando ou simulando arrays bidimensionais em PL/SQL

declare

  type t_features is table of varchar(100) index by pls_integer;
  type t_car_rec is record
  (
    make varchar2(50),
    model varchar2(50),
    features t_features
  );

  type t_car_tab is table of t_car_rec index by pls_integer;
  car_tab t_car_tab;

  procedure show_detail is
    car_idx pls_integer;
    features_idx pls_integer;
  begin
    car_idx := car_tab.first;
    loop
      exit when car_idx is null;
      dbms_output.put_line('Details for ' || car_tab(car_idx).make || ' ' || car_tab(car_idx).model);

      features_idx := car_tab(car_idx).features.first;
      loop
        exit when features_idx is null;
        dbms_output.put_line('   =>' || car_tab(car_idx).features(features_idx));

        features_idx := car_tab(car_idx).features.next(features_idx);
      end loop;

      car_idx := car_tab.next(car_idx);
    end loop;
  end;

begin

  -- using sequential index values
  car_tab(1).make := 'Ferrari';
  car_tab(1).model := 'Testarossa';
  car_tab(1).features(1) := 'Fast';
  car_tab(1).features(2) := 'Looks cool';
  car_tab(1).features(3) := 'Expensive';

  -- using random index values (sparse)
  car_tab(2).make := 'Acura';
  car_tab(2).model := 'TSX';
  car_tab(2).features(14) := 'Small';
  car_tab(2).features(200) := 'Good MPG';
  car_tab(2).features(36) := 'Inexpensive';

  show_detail;

end;

Details for Ferrari Testarossa
  =>Fast
  =>Looks cool
  =>Expensive 
Details for Acura TSX
  =>Small
  =>Inexpensive
  =>Good MPG