An Optimised Flow for Futures: From Theory to Practice

TL;DR

This paper demonstrates that data-flow explicit futures are a powerful primitive for concurrent programming, enabling optimizations like the 'forward' construct, with proven equivalence and practical performance benefits.

Contribution

It proves the equivalence between 'forward' and 'return' in data-flow futures, and provides a library implementation showing practical advantages.

Findings

Data-flow explicit futures unify nested futures into plain futures.

The 'forward' construct can be optimized as a language primitive.

Benchmarks confirm the approach's performance viability.

Abstract

A future is an entity representing the result of an ongoing computation. A synchronisation with a "get" operation blocks the caller until the computation is over, to return the corresponding value. When a computation in charge of fulfilling a future delegates part of its processing to another task, mainstream languages return nested futures, and several "get" operations are needed to retrieve the computed value (we call such futures "control-flow futures"). Several approaches were proposed to tackle this issues: the "forward" construct, that allows the programmer to make delegation explicit and avoid nested futures, and "data-flow explicit futures" which natively collapse nested futures into plain futures. This paper supports the claim that data-flow explicit futures form a powerful set of language primitives, on top of which other approaches can be built. We prove the equivalence, in the context of data-flow explicit futures, between the "forward" construct and classical "return" from functions. The proof relies on a branching bisimulation between a program using "forward" and its "return" counterpart. This result allows language designers to consider "forward" as an optimisation directive rather than as a language primitive. Following the principles of the Godot system, we provide a library implementation of control-flow futures, based on data-flow explicit futures implemented in the compiler. This small library supports the claim that the implementation of classical futures based on data-flow ones is easier than the opposite. Our benchmarks show the viability of the approach from a performance point of view.