A domain-specific dialect for financial-economic calculations using reactive programming

Abstract

This thesis investigates to what extent reactive programming is suited for the implementation of systems that support banks in providing financial-economic advice to their private customers. To this end, we built a small such system in Hotdrink, as a proof-of-concept. This small system is a simplification of a much larger system that has been developed by Delfi Data, and which is overdue to be rewritten from scratch. Our system was evaluated on the commercial criteria readability, maintainability, extensibility, and efficiency. It turned out that readability and maintainability were quite high, even allowing nondevelopers to understand most of the code. However, extensibility is limited for financial calculations in general and especially for parallelization and encapsulation. With regard to commercialization, the solution was too immature to be evaluated on efficiency. Our conclusion is that reactive programming alone is too restrictive for solving financialeconomic problems. Notwithstanding the above, as a by-product we developed a prototype which sourceto-source compiles our new high-level domain-specific JavaScript dialect to reactive calculations.