High Reliability Component-Based Software Engineering

本研究的主要目标是：1.克服软件构件及其组合的结构复杂性；为了克服软件构件及其组合的行为复杂性，我们实现了以下目标：1.软件构件及其组合的行为复杂性。通过创建支持组件清晰而灵活的模块化的模型和语言并基于这些组件构建系统来解决结构复杂性。通过将形式化的验证技术应用于组件的行为，并设计方法来证明它们满足期望的属性，不会遇到失败或带来不良现象，以解决行为复杂性问题。作为最终产品，我们将在四年半的项目中开发的方法和工具集成到一个集成的软件开发环境中。研究产品的一些特点如下。-我们提出了一个名为Epsilon的基于角色的模型和实现该模型的语言EpsilonJ，它支持定义和组合组件的灵活体系结构。-我们开发了一个基本的面向方面的模型框架来比较和分析各种面向方面的编程语言。-为了确保性能，我们设计了一种自动生成适合于部署组件的环境的代码的方法。-我们提出了一种新的系统化的方法来编织在软件开发的早期阶段设计的抽象方面。-我们提出了一个支持相互递归类的新型系统。

The major objectives of this research were:1. to conquer the structural complexity of software components and their composites, and2. to conquer the behavioral complexity of software components and their composites.We achieved these objectives as follows.1. The structural complexity is addressed by creating models and languages that support clear and flexible modularization of components and constructing systems based on those components.2. The behavioral complexity is addressed by applying formal validation techniques to behaviors of components and devising methods for proving that they satisfy desirable properties and will not run into failure or bring undesirable phenomena.As a final product, we integrated methods and tools developed during the four and half year project under an integrated software development environment. Some characteristics of the research product are as listed below.-We proposed a role based model named Epsilon and a language to implement the model, EpsilonJ that supports a flexible architecture defining and combining components.-We developed a fundamental aspect oriented model framework to compare and analyze a wide variety of aspect oriented programming languages.-To ensure performance, we devised a method for automatically generating code that fits to the environment where components are deployed.-We proposed a new systematic method for weaving in abstract aspects designed at the earlier stage of software development.-We proposed a new type system that supports mutually recursive classes.