Proving Deadlock Freedom in Component-Based Programming
Proving Deadlock Freedom in Component-Based Programming
复制标题
证明基于组件的编程中的死锁自由度
DOI：
10.1007/3-540-45314-8_6
复制
发表时间：
2001
影响因子：
0.8
通讯作者：
Sebastián Uchitel
中科院分区：
文献类型：
--
作者：
P. Inverardi;Sebastián Uchitel
新兴的技术，如商业现货产品（COTS）和组件集成框架，如CORBA和COM正在改变软件的生产方式。分布式应用程序被设计为一组自治的、解耦的组件，允许基于COTS集成的快速开发，并简化了科普底层环境动态所需的架构更改。尽管集成技术和开发技术假设了相对简单的体系结构，但它们面临着一个关键问题：构件集成，现有的动态集成错误检测技术都是基于组合系统的行为分析，存在严重的空间复杂性问题。在这项工作中，我们提出了一个更广泛的概念，组件语义的基础上的假设和方法，证明死锁自由在一个基于组件的设置。我们的目标是防止和检测这些错误在组件的编程设置在一个组件明智的方式。我们的目标是寻找有效的方法，即使以不完整为代价，也可以扩展到真实的大小应用程序，而不是许多现有的方法，尽管理论上完整，但在实践中可能会失败。
Emerging technologies such as commercial off-the-shelf products (COTS) and component integration frameworks such as CORBA and COM are changing the way software is produced. Distributed applications are being designed as sets of autonomous, decoupled components, allowing rapid development based on integration of COTS and simplifying architectural changes required to cope with the dynamics of the underlying environment. Although integration technologies and development techniques assume rather simple architectural contexts, they face a critical problem: Component integration.So far existing techniques for detecting dynamic integration errors are based on behavioural analysis of the composed system and have serious space complexity problems. In this work we propose a broader notion of component semantics based on assumptions and a method for proving deadlock freedom in a component-based setting. Our goal is to prevent and detect these errors in component based programming settings in a component-wise fashion. We aim for effective methods that can scale to real size applications even at the price of incompleteness as opposed to many existing methods that although theoretically complete might fail in practice.