Coupled evolution of model-driven spreadsheets

Abstract

Spreadsheets are increasingly used as programming languages, in the construction of large and complex systems. The fact is that spreadsheets, being a highly flexible framework, lack important programming language features such as abstraction or encapsulation. This flexibility, however, comes with a price: spreadsheets are populated with significant amounts of errors. One of the approaches that try to overcome this problem advocates the use of model-driven spreadsheet development: a spreadsheet model is defined, from which a concrete spreadsheet is generated. Although this approach has been proved effective in other contexts, still it needs to accommodate for future evolution of both the model and its instance, so that they remain synchronized at all moments. In this paper, we propose a pair of transformation sets, one working at the model level and the other at the instance level, such that each transformation in one set is related to a transformation in the other set. With our approach, we ensure model/data compliance while allowing for model and data evolution.