CONCEPT Cited by 1 source
Eventual readiness¶
Eventual readiness is the client-facing form of an eventually consistent dependency becoming usable: identity, permissions, indexes, or newly written data exist, but are not yet visible on every path the client needs. Instead of presenting readiness as a Boolean and treating the temporary interval as an error, the application models a converging state with safe degraded behavior and a ready state with authoritative live reads.
Why a Boolean fails¶
A successful login or write often confirms only one subsystem. A later permission check, query index, or read replica can lag. If a client treats “not yet visible” as “does not exist,” it creates empty screens, duplicate setup requests, or an apparent reversal of an accepted action. The user experiences a normal convergence window as a product failure.
Safe behavior during convergence¶
- Render cacheable or bundled data that does not require newly propagated authorization.
- Persist user intent before relying on the remote write path.
- Validate returned content/shape rather than only HTTP success status.
- Retry delivery in the background with bounded, observable behavior.
- Keep acknowledged local state until a live authoritative read verifies it.
The application must identify which data is safe to show while convergence proceeds; this is not permission to show stale or private data indiscriminately.
Seen in¶
- sources/2026-08-07-atlassian-building-a-real-time-pwa-on-atlassians-own-stack — Atlassian's event PWA waits for account provisioning and Jira permission propagation with a readiness state machine. In the converging state it renders safe event data, captures intent locally, and retries delivery; in the ready state live reads become authoritative. The same recovery path handles later degradation. See patterns/eventual-readiness-state-machine.