Industrial Experience with the Migration of Legacy Models using a DSL
Mathijs Schuts, Jozef Hooman, Paul M. Tielemans · 2018
Software departments of companies that exist for several decades often have to deal with legacy models. Important business assets have been modelled with tools that are no longer preferred within the company. Manually remodelling these models with a new tool would be too costly. In this paper, we describe an approach to migrate from Rhapsody models to models of another tool. To perform the migration, we created a Domain Specific Language (DSL) that accepts Rhapsody models as instances. A generator of this DSL can then produces model instances for the new tool. To get confidence in the transformation in a pragmatic way, we applied a combination of model learning and equivalence checking. Learning has been applied to both the source code generated by Rhapsody and the code generated by the new tool. The resulting models are compared using equivalence checking.