A major problem for writing extensible software arises when recursively defined datatypes and operations on these types have to be extended simultaneously without modifying existing code. This paper introduces Extensible Algebraic Datatypes with Defaults which promote a simple programming pattern to solve this well known problem. We show that it is possible to encode extensible algebraic datatypes in an object-oriented language, using a new design pattern for extensible visitors. Extensible algebraic datatypes have been successfully applied in the implementation of an extensible Java compiler. Our technique allows for the reuse of existing components in compiler extensions without the need for any adaptations.

New extensions to programming languages are constantly being proposed. But implementing these extensions usually turns out to be a very di#cult and expensive task, since conventional compilers often lack extensibility and reusability. In this paper we present some fundamental techniques to implement extensible compilers in an object-oriented language. For being able to implement extensible compiler passes, we introduce an extensible form of algebraic datatypes. Our extensible algebraic datatypes with defaults yield a simple programming protocol for implementing extensible and reusable compiler passes in a functional style. We propose an architectural design pattern ContextComponent which is specifically targeted towards building extensible, hierarchically composed systems. Our software architecture for extensible compilers combines the use of algebraic types, known from functional languages, with this object-oriented design pattern. We show that this approach enables us to extend existing compilers flexibly without modifying any source code. Our techniques have been successfully applied in the implementation of the extensible Java compiler JaCo.