Actor reflection without meta-objects

Abstract

We consider how reflection should be introduced into an actor language, where reflection is a mechanism for allowing a program to have access to the data structures of its own processor. Rather than introducing meta-objects as in the previous work, we propose to introduce reflection through two kinds of special messages: reifying and reflecting messages. We show that the fifll range of reflective programming in an actor language is possible without introducing meta-objects, and argue that our approach provides a more uniform interface for actors than the meta-object approach. All the examples in this paper have been tested on our prototype actor system.