Goal-driven combination of software comprehension approaches for component based development

Abstract

This paper reports on our approaches to combine various software comprehension techniques (and technologies) in order to establish con dence whether a given reusable component satis es the needs of the intended reuse situation.Some parts of the problem we are addressing result from di erences in knowledge representation about a component depending on whether this component i s a w ell documented in-house development, some externally built componentry, or a COTS-component. KeywordsProgram comprehension, software visualization, cognitive models, speci cation animation, trace analysis MOTIVATIONWhile the issue of building software from building blocks 12, 15] shifts from using classical reusable building blocks to using o -the-shelf components, modern software technology supports software development on the basis of non-trivial componentry. However, one of the key issues causing the Not-Invented-Here syndrome 29] remains: How can developers be sure that the component they plan to use in their new construction venture meets the expectations placed into it.As long as reuse is con ned to domain-speci c in-house reuse, the question on whether one can trust into a component p r o vided by colleagues is relatively benign. It can be resolved informally and nal con dence might be established by means of test data. As soon as reuse transcends organizational boundaries { and with COTS integration, this is the normal situation {, this informal trust (whether it was ever justi ed or not is here not the issue) is lost. Just testing is insu cient to re-establish it. Speci c mechanisms have t o be devised in order to compensate for the loss of in-depth informal information.