previous section are rooted in the failure of most current programming languages and methodologies to recognize the identification and management of dependencies among software components as a design problem in its own right. This shortcoming translates to inadequate support for making interconnection assumptions embedded inside components more explicit, inability to localize information about interconnection protocols, and a lack of theories and systematic taxonomies of software interconnection relationships and ways of managing them.As a response to this situation, this chapter proposes a new perspective for representing and implementing software systems. Unlike current practice, this perspective emphasizes the explicit representation and management of dependencies among software activities as distinct entities.The perspective is based on the ideas of coordination theory (Malone and Crowston 1994) and the Process Handbook project (Dellarocas et al. 1994; Malone et al. 1993). In accordance with Malone and Crowston (1994), we define coordination as the act of managing dependencies among activities. In this case the activities we are concerned with are software components. We will also use the term coordination process or protocol to describe the additional code introduced into a software system as a result of managing some dependency. These definitions lead to the principles of our coordination perspective on software system design which can be stated as follows:
Explicitly represent software dependencies. Software systems should be described using representations that clearly separate the core functional pieces of an application from their interdependencies, providing distinct abstractions for each.
Build design handbooks of component integration. The field knowledge on component integration should be organized in systematic taxonomies that provide guidance to designers and facilitate the generation of new knowledge. Such taxonomies will catalog the most common kinds of interconnection relationships encountered in practice. For each relationship, they will contain sets of alternative coordination protocols for managing it. In that way they can form the basis for design handbooks of component integration, similar to the well-established handbooks that assist design in more mature engineering disciplines.
The long-term goal of this research is to develop concrete software development tools and methodologies based on the principles stated above, and to demonstrate that such methodologies provide practical benefits in the initial development, maintenance, and reuse of software systems. A crucial part of our efforts revolves around the definition of useful taxonomies of dependencies and coordination protocols and the organization of that knowledge in on-line repositories that can then assist or even automate the integration of the different parts of a system into a coherent whole.
لطفا منتظر باشید ...
