First Class Futures: a Study of Update Strategies

Futures enable an efficient and easy to use programming paradigm for distributed applications. A natural way to benefit from distribution is to perform asynchronous invocations to methods or services. Upon invocation, a request is en-queued at the destination side and the caller can continue its execution. But a question remains: ``what if one wants to manipulate the result of an asynchronous invocation?'' First-class futures provide a transparent and easy-to-program answer: a future acts as the placeholder for the result of an asynchronous invocation and can be safely transmitted between processes while its result is not needed. Synchronization occurs automatically upon an access requiring the result. As references to futures disseminate, a strategy is necessary to propagate the result of each request to the processes that need it. This report studies the efficient transmission of results: it presents three main strategies in a semi-formal manner, and provides a cost analysis with some experiments to determine the efficiency of each strategy.