University of Hertfordshire

By the same authors

A Linear Decomposition of Multiparty Sessions for Safe Distributed Programming

Research output: Contribution to conferencePaper

  • Alceste Scalas
  • Ornela Dardha
  • Raymond Hu
  • Nobuko Yoshida
View graph of relations
Original languageEnglish
Pages24:1-24:31
Number of pages31
DOIs
Publication statusPublished - 16 Jun 2017
Event31st European Conference on Object-Oriented Programming (ECOOP 2017). - Universitat Politècnica de Catalunya (UPC), , Barcelona, Spain
Duration: 18 Jun 201723 Jun 2017
https://2017.ecoop.org/

Conference

Conference31st European Conference on Object-Oriented Programming (ECOOP 2017).
CountrySpain
CityBarcelona
Period18/06/1723/06/17
Internet address

Abstract

Multiparty Session Types (MPST) is a typing discipline for message-passing distributed processes that can ensure properties such as absence of communication errors and deadlocks, and protocol conformance. Can MPST provide a theoretical foundation for concurrent and distributed programming in "mainstream" languages? We address this problem by (1) developing the first encoding of a full-fledged multiparty session pi-calculus into linear pi-calculus, and (2) using the encoding as the foundation of a practical toolchain for safe multiparty programming in Scala. Our encoding is type-preserving and operationally sound and complete. Crucially, it keeps the distributed choreographic nature of MPST, illuminating that the safety properties of multiparty sessions can be precisely represented with a decomposition into binary linear channels. Previous works have only studied the relation between (limited) multiparty and binary sessions via centralised orchestration means. We exploit these results to implement an automated generation of Scala APIs for multiparty sessions, abstracting existing libraries for binary communication channels. This allows multiparty systems to be safely implemented over binary message transports, as commonly found in practice. Our implementation is the first to support distributed multiparty delegation: our encoding yields it for free, via existing mechanisms for binary delegation.

Notes

© Alceste Scalas, Ornela Dardha, Raymond Hu, and Nobuko Yoshida; licensed under Creative Commons License CC-BY

ID: 17769763