Static Application-Level Race Detection in STM Haskell using Contracts

Romain Demeyer, Wim Vanhoof

Research output: Contribution in Book/Catalog/Report/Conference proceedingConference contribution

15 Downloads (Pure)

Abstract

Writing concurrent programs is a hard task, even when using high-level synchronization primitives such as transactional memories together with a functional language with well-controlled side-effects such as Haskell, because the interferences generated by the processes to each other can occur at different levels and in a very subtle way. The problem occurs when a thread leaves or exposes the shared data in an inconsistent state with respect to the application logic or the real meaning of the data. In this paper, we propose to associate contracts to transactions and we define a program transformation that makes it possible to extend static contract checking in the context of STM Haskell. As a result, we are able to check statically that each transaction of a STM Haskell program handles the shared data in a such way that a given consistency property, expressed in the form of a user-defined boolean function, is preserved. This ensures that bad interference will not occur during the execution of the concurrent program.
Original languageEnglish
Title of host publicationElectronic Proceedings in Theoretical Computer Science EPTCS
PublisherOpen Publishing Association
DOIs
Publication statusPublished - 10 Dec 2013
Event5th Workshop on Programming Language Approaches to Concurrency and Communication-cEntric Software (PLACES 2013) - Roma, Italy
Duration: 23 Mar 2012 → …

Publication series

NameEPTCS
PublisherOpen Publishing Association

Symposium

Symposium5th Workshop on Programming Language Approaches to Concurrency and Communication-cEntric Software (PLACES 2013)
Country/TerritoryItaly
CityRoma
Period23/03/12 → …

Keywords

  • cs.LO
  • cs.DC
  • cs.PL

Fingerprint

Dive into the research topics of 'Static Application-Level Race Detection in STM Haskell using Contracts'. Together they form a unique fingerprint.

Cite this