Using domain-specific languages for the realization of component composition

Abstract

In recent years, component-based development has evolved to one of the key technologies in software engineering, because it provides a promising way to deal with large scale software developments. Due to this, the realization of component interactions has become an important task while implementing a system being assembled from (existing) components. Scripting languages, like Perl, Tcl, Unix-Shell, are often used for implementing this so-called glue code, because they provide a flexible way to process string-based input, the most common data structures used for component interactions. However, often it turns out that the algorithms of the component interactions are too sophisticated to be ad equately expressed in this kind of languages. In this paper, we propose the use of language technology for that purpose: the strings passed be tween the components are treated as sentences in specialized component interaction languages (CIL). The syntax of such a language defines the input format of the components to be interconnected, and the semantics represents the functionality of the glue code. The appropriateness of this approach depends on the methodology and support functionality available for designing these languages. We report on the positive experiences using Montages as methodology. We will also describe the support functionality of the Gem-Mex tool containing its graphical animation and debugging facilities, which can be used as vehicle for the comprehension of the interplay between the components of the overall system.