An alternative to software process languages

We argue that years of research into the characteristics needed in an "ideal" software process definition language have only demonstrated that such a language, if one were to exist, would be impossibly large and complicated. We suggest that it would be most difficult to write clear and effective process definitions in such a language. Thus, the prospects for continuing with this line of research seem bleak. We suggest that, instead we should model software processes using a variety of modelling formalisms, and we should implement software process execution systems as distributed software systems. These systems should be composed of a heterogeneous mix of process fragments, written in different languages designed to support the gamut of different specialized tasks that we now understand to be necessary.