representation-decoupling-spans-module-and-architecture-scales

Status: IN

Representation decoupling operates at two architectural scales through the same principle: within modules, input representations are separated from storage representations (dataclasses vs dicts, coordinator writes vs raw replicas); across the architecture, logical semantics are separated from physical state (offsets survive partition trimming, hash positions decouple from topology, delete markers decouple from data removal) — both enabling independent evolution of interfaces and implementations at their respective granularities.

Justifications

Representation decoupling at module scale and architecture scale are the same principle at different granularities

Depends on (SL): input-and-storage-representations-are-decoupled, logical-physical-separation-spans-topology-storage-and-deletion

JSON