A Modular Event-Driven Messaging Framework for Web and Mobile Applications: Unifying REST and WebSocket Access over Declared Channel Semantics

Abstract

Modern web and mobile applications are event-driven at the experience level, feeds refresh, presence changes, collaborative edits land, orders progress, while remaining request-driven at the architecture level, built on REST endpoints with real-time behavior retrofitted through ad hoc WebSocket handlers, duplicated delivery logic, and reconnection code rewritten per feature. The costs are structural: state divergence between what requests return and what events pushed, no uniform semantics for ordering, resumption, or backpressure, and mobile clients whose offline behavior is an afterthought. This paper proposes a modular event-driven messaging framework for web and mobile applications that treats REST and WebSocket as two disciplined faces of one event-oriented core. The framework defines a channel abstraction, named, typed event streams with declared delivery semantics, per-channel ordering keys, and retention windows, produced by application services into an event core and consumed uniformly over three access modes: request-response snapshots over REST, live subscription over WebSocket with sequence numbering and snapshot-plus-delta resumption, and hybrid read-then-subscribe that guarantees gapless transition from snapshot to stream. Modularity is achieved through a layered contract: transport adapters, session and subscription management, channel routing with per-channel policy, and application bindings are separately replaceable, with server-sent events and polling adapters degrading gracefully where full-duplex transport is unavailable. Client libraries specify offline queueing, idempotent command submission, and convergent local caches for mobile conditions. The framework’s semantics, delivery guarantees per channel class, reconnection contracts, and backpressure policies, are stated precisely, and an evaluation plan measures latency, resumption correctness under churn, server fan-out capacity, and developer-facing modularity through feature-implementation studies. Applications, limitations, and migration pathways from REST-only systems are analyzed.