Caching

Out of the box EquinoxJS supports an in-memory LRU cache. This cache is based on the lru-cache library. The default configuration allows 1,000,000 items to live in the cache. Since it's impossible to know the size of a JS object in bytes at runtime we cannot provide a sensible default for the cache that limits the actual memory usage of your process. You'll need to fine tune it to your actual requirements. As an example if you have 1 million states, each of which is 50 bytes that's 50MB of memory.

The cache doubles as a concurrency limiter on stream loads and guarantees that for each stream at most one load is happening concurrently.

For performance reasons we do not offer TTL functionality on the cache. Doing so with the guarantees we'd want is not feasible with the underlying cache library. In particular, the underlying library will not pre-emptively evict expired items from the cache and as such your items would remain in the cache until requested again, at which point they'd be evicted. As such caching in EquinoxJS is a binary on/off setting.

NoCaching

import { CachingStrategy } from "@equinox-js/core"
const strategy = CachingStrategy.NoCache()

No caching is applied. This triggers a fresh load of the entire stream for every interaction with the store.

Cache(cache, ?prefix)

import { CachingStrategy, MemoryCache } from "@equinox-js/core"
const cache = new MemoryCache()
const strategy = CachingStrategy.Cache(cache)

Retain a single state per stream_name in the supplied cache.

Prefix is used to segregate multiple folds per stream when they are stored in the cache.