A contract-based adaptive policy framework for progressive web applications under unstable network conditions

Abstract

The paper examines the transition from an application-specific adaptive-policy method to a reusable framework for progressive web applications in which unstable network connectivity affects data retrieval, content volume, and the completion of write operations. The previous method was implemented for a single application and therefore did not allow evaluation of core reuse, consistent policy activation across the window context and the Service Worker, or the correctness of deferred operations under failures. A contract-based software framework is proposed that combines a shared core with an application specification containing an operation registry, a policy, adapters, and backend capabilities. The experimental evaluation included comparison with a recorded behavioral baseline, controlled fault injection, and verification of core reuse across two web applications. Complete decision agreement and identical browser outcomes were obtained for all pairs in the main experiment, while timing overhead remained within the protocol-defined limits. All 54 fault-matrix executions produced the expected outcomes, and repeated delivery under an atomic backend contract caused no more than one state change per idempotency key. Both web applications used the same core and passed the shared checks. The results confirm the reusability of the framework; however, transferring this conclusion to other backend implementations requires separate validation.