Perspectives of Model Transformation Reuse

Model Transformations have been called the “heart and soul” of Model-Driven software development. However, they take a lot of effort to build, verify, analyze, and debug. It is thus imperative to develop good reuse strategies that address issues specific to model transformations. Some of the effective reuse strategies are adopted from other domains, specifically, programming languages. Others are custom developed for models. In this paper, we survey techiques from both categories. Specifically, we present two techniques adoped from the PL world: subtyping and mapping, and then two techniques, lifting and aggregating, that are novel in the modeling world. Subtyping is a way to reuse a transformation for different but similar input modelling languages. Mapping a transformation designed for single models reuses it for model collections, such as megamodels. Lifting a transformation reuses it for aggregate representations of models, such as product lines. Aggregating reuses both transformation fragments (during transformation creation) and partial execution results (during transformation execution) across multiple transformations. We then point to potential new directions for research in reuse that draw on the strengths of the programming and the modeling worlds.