Modifying the forrest-tomlin and saunders updates for linear programming problems with variable upper bounds