Consistency management via a combination of triple graph grammars and linear programming

Abstract

Consistency management is an important task in the context of Domain-Specific Language (DSL) development. It involves operations such as program (model) transformation, synchronisation, integration, and consistency checking, which are all tasks required to enable concurrent engineering using multiple DSLs. Even though consistency management is a well-researched topic, existing approaches either implement a fixed strategy for each consistency management operation, or do not scale for large models. This has been criticised in the literature, as practical applications require not only reasonable scalability with model size, but also unite multiple consistency management tasks within one tool. To raise the adaptability of such a tool to an appropriate level, a uniform way of performing these tasks is a desirable goal. In this paper, we propose an approach to consistency management that leverages a synergetic combination of Triple Graph Grammars and Integer Linear Programming. By modelling consistency management as an optimisation problem with a configurable objective function, we are able to uniformly address a wide range of consistency management operations. We show that our approach scales acceptably in practice, while still guaranteeing that a consistent solution is found if and only if one exists.