Résumé
With prototype-based languages, concretization and abstraction are unified into a single concept a prototype. Prototype-based languages are based on a simple set of principles: object-centered representation, dynamic reshape of objects, cloning and possibly message delegation. However, they all differ in the interpretation and combination of these principles. Therefore there is a need to compare and understand. In this paper we present Prototalk, a research and teaching vehicle to understand, implement and compare prototype-based languages. Prototalk is a framework that offers a predefined set of language data structures and mechanisms that can be composed and extended to generate various prototype-based language interpreters. It presents a classification of languages based on different mechanisms in an operational manner.