Dependency resolution (DR) uses a component's explicitly declared requirements to calculate systems where all dependencies are satisfied. There can be many configurations to choose from when resolving dependencies. DR should aim to identify and return an optimal component configuration. This becomes a significant challenge when diverse and sometimes conflicting criteria such as user preferences, contextual constraints and functional requirements must be considered. In this paper we present a framework in which to represent and compose such criteria. This is achieved by defining criteria as ranking systems over complete lattices and composing them in different ways. We present a depth first branch and bound algorithm for this framework, and an example problem that demonstrates the frameworks application. The presented framework will enable the formal definition and composition of criteria to optimise dependency resolution.

Software components are encapsulated units of execution which express dependencies through explicitly stated requirements and capabilities. These are used within component repositories to resolve sub-systems of deployable components. The main problem with this dependency resolution is not the lack of available solutions but the excess of them. Within current repositories the returned solution may not be optimal for the composers requirements and context. Improving dependency resolved solutions from repositories using soft constraints with validation and verification is the topic of our Ph.D. and discussed within this paper.