SoD Collaborative Research: Constraint-based Architecture Evaluation

这是一个合作研究项目，包括德克萨斯大学奥斯汀分校的德韦恩·佩里、詹姆斯·布朗和萨夫拉兹·库尔希德(奖项#0438967)，以及麻省理工学院的丹尼尔·杰克逊(奖项#0438897)。这个项目的目的是开发新的语言、方法、分析和工具，用于“规范的”软件体系结构描述，帮助复杂软件系统的设计者在实现之前表达和探索他们的体系结构承诺，并支持共享体系结构特征的系统系列。使用了两种支撑技术：Jackson的合金约束语言和分析器，以及Browne的组合编译器。扩展了合金的底层逻辑，添加了新的分析功能，并将合金系统打包为软件组件。组合编译器被扩展为利用约束和属性规范来支持约束和性能分析。在这些技术的基础上，建立了软件体系结构的抽象模型，定义了体系结构约束描述语言，定义了组合规则和评估规则，构建了一个集成的基于约束的体系结构评估工具集。首先通过简单的具体实例对研究进行评估，然后结合现实的案例研究，根据需要对研究进行修改和扩展，以满足总体评估目标。本项目的结果将在建筑研究、建筑教育和建筑支持技术方面产生重大而广泛的影响。此外，预计该技术将使人们能够更深入地了解设计权衡，将提高设计质量，并将显著改善组件组装过程。项目网站(http://www.ece.utexas.edu/~perry/work/projects/nsf-cbae/)将提供对所产生的技术和工具的访问。

This is a collaborative research project including Dewayne Perry, James Brown and Sarfraz Khurshid at the University of Texas at Austin (award # 0438967), and Daniel Jackson at the Massachusetts Institute of Technology (award # 0438897). The aim of this project is to develop new languages, methods, analyses and tools for 'prescriptive' software architecture descriptions, to help the designers of complex software systems express and explore their architectural commitments prior to implementation, and to support families of systems that share architectural features. Two underpinning technologies are used: Jackson's Alloy constraint language and analyzer, and Browne's compositional compiler. Alloy's underlying logic is expanded, new analysis capabilities are added, and the Alloy system is packaged as a software component. The compositional compiler is extended to utilize constraints and property specifications in order to support both constraint and performance analyses. On the basis of these technologies, an abstract model of software architecture is created, an architecture constraint specification language is defined, the rules of composition and evaluation are defined, and an integrated tool suite for constraint-based architectural evaluation is constructed. The research is evaluated first by simple specific examples and then with realistic case studies, revising and extending the research as needed to satisfy the overall evaluation goals.The results of this project will have a significant and broad impact in terms architectural research, architecture education and architecture support technology. Further, it is expected that the technology will enable deeper understandings of design trade-offs, will improve design quality, and will significantly improve the component composition process. The project Web site (http://www.ece.utexas.edu/~perry/work/projects/nsf-cbae/) will provide access to the resulting techniques and tools.