Law-Governed Systems

研究了一种处理大型软件系统的新方法。这一方法的中心概念是法治制度(LGS)。非正式地说，法治制度可以用以下等式来描述：法治制度=制度+法律如果法律是关于制度结构和运作的一套明确和严格执行的规则，关于制度的使用和修改的方式，以及法律本身允许的修改。换句话说，在LGS方法下，传统的制度概念得到了一个新的组成部分，称为制度法则，它既规范着制度的运作，也规范着制度的演变。它的目的是构建一个称为达尔文的编程环境来支持这种方法。法治系统的概念似乎有广泛的应用，从控制程序的运行--传统上是编程语言的领域--到控制大型系统的开发过程--传统上是软件开发环境和管理学科的领域。对软件工程的具体贡献预计包括：通用和灵活的模块互连方案，异常处理的全局处理，以统一方式支持各种继承方案的面向对象的语言，监视和审计程序的操作和演变的能力等。也许最有趣的，和潜在的有用的，所提出的方法的预期结果是，一个进化系统可以被制成一个单一的“有机体”，在其整个进化生命周期中表现出一定程度的可预测性和连贯性。这一结果在传统技术中是无与伦比的，对于必须为企业提供长期、几乎不可中断的支持的系统来说，这一结果尤其重要。

A new approach for dealing with large scale software system is investigated. The central concept of this approach is that of a law- governed system (LGS). Informally, a law-governed system can be described by means of the following equation law-governed system = system + law where the law is an explicit and strictly enforced set of rules about the structure and operation of the system, about the way the system is to be used and modified, and about the allowable modifications in the law itself. In other words, under the LGS approach, the conventional notion of a system is augmented by a new component called the law of the system, which regulates both its operation and evolution. It is intended to build a programming environment, called Darwin, to support this approach. The concept of law-governed system appears to have a wide range of applications, ranging from controlling the operation of programs -- traditionally the domain of programming languages -- to controlling the process of development of large scale systems -- traditionally the domain of software-development environments and managerial disciplines. Specific contributions to software engineering are expected to include: general and flexible module-interconnection schemes, global treatment of exception handling, an object-oriented language which supports a wide variety of inheritance schemes in a unified manner, ability to monitor and audit the operation and the evolution of programs, etc. Perhaps the most interesting, and potentially useful, result expected from the proposed approach is that an evolving system could be made into a single "organism" which would exhibit a degree of predictability and coherency throughout its evolutionary lifetime. This result, which has no counterpart in conventional technology, is particularly important for systems that must provide long term, virtually uninterruptable, support for an enterprise.