[Excerpt from the introduction] The spreading of Distributed Memory Parallel Computers (DMPCs) is hampered by the fact that writing or porting application programs to such architectures is a difficult, time-consuming and error-prone task. Nowadays software environments for commercially available DMPCs mainly consist of libraries of routines to handle communications between processes. We believe that the reuse of carefully designed software components could help to manage the complexity of concurrent programming. Our approach aims at embedding the scalable data parallelism programming model in an OOL. This approach is orthogonal to those where objects can be made active and invocations of methods result in actual message passing communications (sometimes referred to as functional parallelism, because the definition of processes is determined by the decomposition into sub-tasks). We believe that functional parallelism does not allow an easy and efficient mapping of large scale computing algorithms onto the very high number of processors that are made available in a DMPC. [...]