CardyGAn: Tool Support for Cardinality-based Feature Models

Cardinality-based feature models (CFM) constitute a crucial and non-trivial extension to FODA feature models in terms of UML-like feature multiplicities and corresponding cardinality constraints. CFM allow for specifying configuration choices of software systems incorporating multiple instances (copies) of features, e.g., for tailoring customer-specific and even potentially unrestricted application resources. Nevertheless, the improved expressiveness of CFM compared to FODA feature models complicates configuration semantics, including sub-tree cloning and potentially unbounded configuration spaces. As a consequence, entirely novel anomalies might arise such as dead cardinality intervals, false unboundedness, and cardinality gaps, which are not properly treated by recent feature-modeling tools. In this paper, we present comprehensive tool support for assisting specification, validation, and configuration of CFM. Our tool CARDYGAN, therefore, incorporates capabilities for CFM editing, automated CFM validation including anomaly detection based on a combination of ILP and SMT solvers, as well as a CFM configuration engine based on ALLOY.