cache

Read-through caching prevents stale reads by using a getOrSet pattern to fetch data from an async factory on a cache miss, then reusing the cached result until the TTL expires or the key is invalidated after repository writes.

Cache invalidation after writes works by explicitly deleting the affected cache keys following repository updates, ensuring that subsequent read-through requests fetch the latest state from the database instead of returning stale data.

You can select cache providers per environment using environment-driven configuration, typically choosing an in-memory store for local development and Upstash or Redis for Vercel preview and production deployments.

The best way to test caching logic without network calls is using a FakeStore to seed deterministic cache values, allowing you to assert key presence or absence and verify getOrSet miss semantics safely.

Read-through caching requires JSON serialization on both read and write operations to ensure cached values are stored and retrieved consistently across different cache providers.

You should use getOrSet semantics for caching derived query results when you want to automatically execute an async factory to fetch data on a cache miss and reuse the stored result for subsequent hits until invalidation.