An Action-based Model to Handle Cloning and Adaptation in Tabular Data Applications

Many software systems require diverse data gathering and handling through processes that manipulate tabular data, often with a spreadsheet orientation. Variability in tabular data cannot be captured in a complete up-front analysis as everything is done at the final user level. She progressively adapts or clones some tabular data organized to conduct a process. Consequently these organized data are constantly both a final usable product and a potential candidate for cloning. This huge diversity, the high frequency of their evolution over time, and the intrinsic need to use cloning lead naturally to the usage of a clone-and-own approach with well-known negative impacts on maintenance and quality. In this paper we advocate that this can be replaced by controlling the clone-and-own process with provenance information that completely captures, at the domain level, the cloning actions and the adaptations applied on a product defining its clones. Each action over the process, its observations, and its data are captured in a complete model through traces of atomic adaptations, complemented with specific derivation and extraction actions. This model enables obtaining the whole history of both data and processes over time, as well as the accountability of variability-related actions. We report on a study showing the relevance of tackled problem in a variability-rich agronomy software of an industrial partner. We also show that a first prototype covers the extracted usage scenarios, from simple and entire cloning to more complex partial cloning.