The CAP theorem reveals a fundamental trade-off between consistency and availability (and, by extension, latency) in systems where network partitions are unavoidable, as in cloud computing and local-first software. Adopting weak consistency can preserve availability but may introduce inconsistencies that compromise application correctness. Replicated data types offer a principled, coordination-free approach to guaranteeing convergence, yet they do not account for application invariants. Existing methods for maintaining invariants in replicated systems either depend on coordination — undermining the benefits of weak consistency — or are limited in scope.

This paper introduces the No-Op framework, a general approach to enforcing consistency without coordination while guaranteeing both convergence and invariant preservation. The central idea is to resolve conflicts among concurrent operations by prioritising one over another according to programmer-defined conflict-resolution policies. This prioritisation reduces the less-preferred operation to a no-side-effect operation, ensuring conflict-free execution. We formalise the model underlying the No-Op framework and present a replication protocol built upon it, along with formal proofs of correctness for both the framework and the protocol. We further demonstrate the framework’s applicability by showing how it supports the design of widely used replicated data types and the preservation of a broad range of application invariants.