Ubiquitous connectivity of web, mobile, and IoT computing platforms has fostered a variety of distributed applications with decentralized state. These applications execute across multiple devices with varying reliability and connectivity. Unfortunately, there is no declarative fault-tolerant programming model for distributed interactive applications with an inherently decentralized system model. We present a novel approach to automating fault tolerance using high-level programming abstractions tailored to the needs of distributed interactive applications. Specifically, we propose a calculus that enables formal reasoning about applications' dataflow within and across individual devices. Our calculus reinterprets the functional reactive programming model to seamlessly integrate its automated state change propagation with automated crash recovery of device-local dataflow and disconnection-tolerant distribution with guaranteed automated eventual consistency semantics based on conflict-free replicated datatypes. As a result, programmers are relieved of handling intricate details of distributing change propagation and coping with distribution failures in the presence of interactivity. We also provides proofs of our claims, an implementation of our calculus, and an empirical evaluation using a common interactive application.

The execution of an application written in a reactive language involves transfer of data and control flow between imperative and reactive abstractions at well-defined points. In a multi-threaded environment, multiple such interactions may execute concurrently, potentially causing data races and event ordering ambiguities. Existing RP languages either disable multi-threading or handle it at the cost of reducing expressiveness or weakening consistency. This paper proposes a model for thread-safe reactive programming (RP) that ensures abort-free strict serializability under concurrency while sacrificing neither expressiveness nor consistency. We also propose an architecture for integrating a corresponding scheduler into the RP language runtime, such that thread-safety is provided "out-of-the-box" to the applications. We show the feasibility of our proposal by providing and evaluating a ready-to-use implementation integrated into the REScala programming language. The scheduling algorithm is formally proven correct. A thorough empirical evaluation shows that reactive applications build on top of it scale with multiple threads, while the scheduler incurs acceptable performance overhead in a single-threaded configuration. The scalability enabled by our scheduler is roughly on-par with that of hand-crafted application-specific locking and better than the scalability enabled by a scheduler using an off-the-shelf software transactional memory library.

Reactive programming improves the design of reactive applications by relocating the logic for managing dependencies between dependent values away from the application logic to the language implementation. Many distributed applications are reactive. Yet, existing change propagation algorithms are not suitable in a distributed setting.