BXtendDSL: A layered framework for bidirectional model transformations combining a declarative and an imperative language
Published in Journal of Systems and Software, 2022
DOI: 10.1016/j.jss.2022.111288
Abstract
Model-driven software development (MDSD) heavily relies on model transformations. While in a strict forward engineering process unidirectional transformations are used, bidirectional transformations are crucial as soon as roundtrip engineering comes into play. In this paper, we present a hybrid language specifically designed to describe bidirectional model transformations. From a declarative transformation specification code is generated which uses our framework for bidirectional and incremental model transformations. A sophisticated code generation mechanism allows for hooking into the generated transformation code at the imperative level to supply behavior that can not be expressed declaratively. A thorough evaluation demonstrates conciseness, expressiveness, and scalability of our approach.