A Difference Calculus for Structured Data

Abstract

Traditional software configuration management tools employ existing Diff tools to identify the difference between intermediate versions of a document such as a source code file. When any new data structure arises, developers have to create a version control tool for this specific structure. This paper introduces a universal approach to the creation of a version control tool for artifacts of any structured data. The approach taken is to develop a difference calculus for arbitrary objects based on their types. This calculus is built on a type system in which types are nested. By constructing types recursively, more structured data can be modelled and the difference between successive versions of objects can be identified.