Transparent First-class Futures and Distributed Components

  • Authors:

  • Antonio Cansado;Ludovic Henrio;Eric Madelaine

  • Affiliations:

  • INRIA Sophia Antipolis, CNRS -- I3S -- Univ. Nice Sophia Antipolis 2004, Route des Lucioles, BP 93, F-06902 Sophia-Antipolis Cedex -- France;INRIA Sophia Antipolis, CNRS -- I3S -- Univ. Nice Sophia Antipolis 2004, Route des Lucioles, BP 93, F-06902 Sophia-Antipolis Cedex -- France;INRIA Sophia Antipolis, CNRS -- I3S -- Univ. Nice Sophia Antipolis 2004, Route des Lucioles, BP 93, F-06902 Sophia-Antipolis Cedex -- France

  • Venue:
  • Electronic Notes in Theoretical Computer Science (ENTCS)
  • Year:
  • 2010

Quantified Score

Hi-index 0.02

Visualization

Abstract

Futures are special kind of values that allow the synchronisation of different processes. Futures are in fact identifiers for promised results of function calls that are still awaited. When the result is necessary for the computation, the process is blocked until the result is returned. We are interested in this paper in transparent first-class futures, and their use within distributed components. We say that futures are transparent if the result is automatically and implicitly awaited upon the first access to the value; and that futures are first-class if they can be transmitted between components as usual objects. Thus, because of the difficulty to identify future objects, analysing the behaviour of components using first-class transparent futures is challenging. This paper contributes with first a static representation for futures, second a means to detect local deadlocks in a component system with first class futures, and finally extensions to interface definitions in order to avoid such deadlocks.