In designing distributed and parallel systems there are several approaches for programming interactions in a multiprocess environment. Usually, these approaches take care only of synchronization or communication in two-party interactions. This paper is concerned with a more general concept: multiparty interactions. In a multiparty interaction, several executing threads somehow "come together" to produce an intermediate and temporary combined state, use this state as a well-defined starting point for some joint activity, and then leave this interaction and continue their separate execution. The concept of multiparty interactions has been investigated by several researchers, but to the best of our knowledge, none have considered how faults in one or more participants of the multiparty interaction can best be dealt with. The goal of this paper is twofold: to show how an existing specification language can be extended in order to allow dependable multiparty interactions (DMIs) to be declared and to present an object-oriented framework for implementing DMIs in distributed systems. To show how our scheme can be used to program a system in which multiparty interactions are more than simple synchronizations or communications, we use a case study based on an industrial production cell model developed by Forschungszentrum Informatik, Karlsruhe, Germany.
翻译:在设计分布式和平行的系统时,有几种在多过程环境中进行编程互动的方法。通常,这些方法只注意双方互动的同步或沟通。本文关注一个更普遍的概念:多党互动。在多党互动中,一些执行线索以某种方式“联合起来”产生一个中间和临时的组合状态,将这个状态作为某种联合活动的明确起点,然后离开这种互动,继续其单独实施。多党互动的概念已经由数位研究人员调查过,但据我们所知,没有任何人考虑过多党互动的一个或多个参与者的缺陷如何能够得到最好的处理。本文的目标有两个方面:说明如何扩展现有的规格语言,以便允许可信赖的多党互动(DMIs)被宣布,并提供一个在分布式系统中实施DMI的面向目标的框架。为了表明我们的计划如何用于规划一个多党互动不仅仅是简单的同步或沟通的系统。我们用的是基于德国卡尔斯鲁希开发的Forschungszentrum Instmatrik的工业生产细胞模型的案例研究。