Relational Expressions for Data Transformation and Computation

Abstract

Separate programming models for data transformation (declarative) and computation (procedural) impact programmer ergonomics, code reusability and database efficiency. To eliminate the necessity for two models or paradigms, we propose a small but high-leverage innovation: the introduction of complete relations into the relational database. Complete relations and the discipline of constraint programming, which concerns them, are founded on the same algebra as relational databases. We claim that by synthesising the relational database of Codd and Date, with the results of the constraint programming community, the relational model holistically offers programmers a single declarative paradigm for both data transformation and computation, reusable code with computations that are indifferent to what is input and what is output, and efficient applications with the query engine optimising and parallelising all levels of data transformation and computation.