The use of frameworks in the context of modern software development promises high levels of reuse of design knowledge as well as of proven implementation artefacts. Unfortunately the expected benefit of an optimisation of time, cost and quality is often not seen in practice. A decisive cause for this can often be found – besides the lack of quality – in the monolithic design of frameworks: the training is too complicated and the possibilities of extension too limited. The presented work addresses this problem with a concept of modular component frameworks. In this case a component framework is understood as a “semi-finished” specification and implementation of an isolated view upon an application system explicitly designed for integration with further component frameworks. This implies an understanding of software development as selection, composition, configuration and instantiation of component frameworks. In this approach, role-based models represent precisely the conceptual understanding of an application domain and, therefore, the premise for integration of several separated views. By the alignment of implementation structures to these abstract models, the integration knowledge from the model layer can widely be reused for the composition of implementation artefacts. By a formal system model the introduced concepts are defined, the various layers of a component framework are related to each other, and, finally, the variants of integration are examined and evaluated. A general application example demonstrates the software development by means of component frameworks and accompanies the practical implementation using modern component technologies. In order to meet the high complexity of modular frameworks, essential aspects of a methodology for design and application of component frameworks are presented. This includes a demand-driven extension of the graphical description technique UML as well as a flexible, extensible process model based on process patterns.     «

The use of frameworks in the context of modern software development promises high levels of reuse of design knowledge as well as of proven implementation artefacts. Unfortunately the expected benefit of an optimisation of time, cost and quality is often not seen in practice. A decisive cause for this can often be found – besides the lack of quality – in the monolithic design of frameworks: the training is too complicated and the possibilities of extension too limited. The presented work addresse...     »