Model Transformation Co-evolution: a Semi-automatic Approach

Model transformations are precious and effortful outcomes of Model Driven Engineering. Their maintainability is affected by changes in not only the transformation requirements but also the associated metamodels. Corrective or perfective metamodel changes certainly percolate the transformation counterparts. Manual migration is cumbersome and error-prone. This paper introduces a semi-automatic approach geared towards ensuring that the transformation does not have syntactic errors. That is, EMOF/Ecore-based metamodel changes are propagated
to transformations so that evolved transformations do not stop working. To this end, we introduce a taxonomy of metamodel changes, we analyze their potential impact on transformations, and present a set of contingency actions to restore the syntactic validity of the transformation rules. This approach, though limited, allows the designer to focus on additive metamodel changes, leaving the automatically resolvable ones to
the transformation assistant. A prototype implemented for ATL serves as a proof of concept.

This publication has not any associated prototype.


University of the basque country