Summary: | We present a simple model of shared data-objects, which extends the abstract data type model to support distributed programming. Our model essentially provides shared address space semantics, rather than message passing semantics, without requiring physical shared memory to be present in the target system. We also propose a new programming language, Orca, based on shared data-objects. A compiler and three different run time systems for Orca exist, which have been in use for over a year now. 1. INTRODUCTION The growing interest in distributed computing systems has resulted in a large number of languages for programming such systems [Bal et al. 1989c]. Many of these languages are oriented towards systems programming and are typically used for distributed operating systems, file servers, and other systems programs. In this paper we will discuss a language designed for implementing distributed user applications. In particular, our language is intended for parallel, high-performance applic.
|