Skip to Main Content
Software modules and components have always played a key role in software engineering, primarily as coarse-grained abstractions that embody the principle of information hiding. The fundamental contribution of component-based software engineering lies in making the black-box property a core requirement. In this paper we point out that this important notion tends to be neglected in component model analyses, and define the black-box property in terms of detailed characteristics which help to evaluate component models from this perspective. A survey of typical representatives of industrial and research component models shows that especially the former ones leave much to be desired, and the paper discusses the consequences of the weak aspects of their design.