Unifying Correctness for Communicating Software

Information / latest news

  • This is a project made possible by a 5-year career grant awarded to Jorge A. Perez by NWO (Netherlands Organization for Scientific Research)

Vacancies

  • A PhD position is available! See here for details and instructions on how to apply. Deadline: July 31, 2018. Early expressions of interest are most welcome.

Summary

In computer science, the correctness problem consists in checking that computer systems behave as intended. This long-standing challenge is fundamental today, as most aspects of society depend on communication-based software systems. These systems rely on message-passing programs, which implement complex protocols often exploiting concurrency. A major current challenge is to make these programs respect their intended protocols when deployed in larger systems.

To meet this challenge, this project aims to develop the first unified theory of correctness for message- passing concurrency. We will focus on so-called behavioral type systems that enforce correct message-passing programs by codifying intended protocols as types. These type systems stand out due to their mathematical foundations and independence from specific programming paradigms. Many behavioral type systems exist, but their precision varies ostensibly. As a result, there are as many notions of correctness as there are behavioral type systems; this makes associated verification toolchains hardly interoperable. To overcome this bottleneck, the unified theory will relate the notions of correctness that existing behavioral type systems enforce.

To achieve this aim, various verification techniques for message-passing programs based on behavioral type systems will be rigorously related by applying a new reference framework that will be developed. A key innovation will be the use of the Curry-Howard correspondence for concurrency, the most principled link between computation and logic, to articulate a unified theory of correctness. These developments will crucially exploit our recent underpinning results and our long track record on comparative expressiveness for concurrency. These foundational results will be validated through case studies and tool prototypes.

This project will deliver a sorely missing foundational reference for many advanced verification techniques that target the correctness problem in message-passing concurrency. In the longer term, the project outcomes will ensure that message-passing programs work seamlessly when deployed in society.