Introduction

The object-oriented paradigm has been adopted by an increasing number of programmers and organizations over the last decade because of its clear advantages in organizing and reusing software components. It would clearly be advantageous to be able to provide static type systems for object-oriented languages that are of the same quality as those available for more standard procedural languages. Unfortunately commercially available object-oriented languages fall far short of that goal. The static type systems of object-oriented languages tend to be either insecure or more inflexible than one might desire. In some cases the rigidity of the type system leads programmers to rely on type casts (sometimes checked at run-time, sometimes not) in order to obtain the expressiveness desired. In other cases, the type systems are too flexible, requiring the run-time system to generate link-time or run-time checks to ensure the integrity of the computation. In this paper we explore the type-checking systems of object-oriented programming languages: examining problems and suggesting solutions.