Type-Safe and Conformance-Preserving Composition and Typing of Metamodels with Co-evolution of Models

Abstract

Model-driven engineering is an approach that has increased in popularity during the course of the last couple of decades. In model-driven engineering models are first-class artefacts, which provides the software engineers with a powerful abstraction for handling complexity and ensuring separation of concerns. This allows engineering software that is easier to understand and reason about than traditional approaches to software engineering, which in turn lowers the costs of development and maintenance.

A key to the success of model-driven engineering is model management operations, e.g. model composition and model migration. However, current approaches for metamodel composition are not type-safe, or address how existing models are impacted when metamodels are composed. Ensuring type-safety and correctness of metamodel composition approaches, with following model migration and model composition, is of great importance.

This thesis describes 1) how the structure and operational semantics of metamodels can be composed and adapted type-safely while ensuring the validity of existing models, 2) how the operational semantics of metamodels can be integrated practically non-intrusively and 3) how metamodels can be typed to support variance and reuse.

Three main results are presented. The first result is a collection of language constructs that realise a type-safe template-based mechanism for composition and adaptation of metamodels, including a framework that migrate and/or compose existing models. The result also includes a framework for formal analysis of these operations. The second result is two mechanisms for integrating the operational semantics of metamodels, as defined by class operations, practically non-intrusively. The third result is a theory for realising metamodel types by means of class nesting, and how this allows substitutability and polymorphism for metamodels as a whole.