@chttp/cache

@chttp/cache

LRU + TTL in-process memory cache with lru-cache-parity semantics, authored in the scriptc-compilable subset — the same source runs under Node and lowers into a native ELF.

Quickstart

pnpm add @chttp/cache

import { Cache } from "@chttp/cache";

const sessions: Cache<string> = new Cache({ max: 10_000, ttl: 60_000 });

sessions.set("sess-1", "uid:42");
sessions.get("sess-1"); // "uid:42" — bumps recency
sessions.get("missing"); // null — misses are never undefined

import { Cache } from "@chttp/cache";

const seen: Cache<number> = new Cache({ max: 1_000 });
seen.set("job-7", 1);

scriptc build src/main.ts -o app

Semantics

• get bumps recency; peek and has don’t. Recency picks the eviction victim when size > max.
• max evicts least-recently-used first (insertion-ordered Map); ttl expires lazily on access — no timers.
• Misses return null; has() distinguishes a stored null from a miss.
• set(key, value, { ttl }) overrides the constructor TTL per entry.
• Non-positive max/ttl throws E_BAD_CACHE_OPTIONS at construction — config fails fast, traffic never does.

Divergences from lru-cache

Bigger than memory

When one process isn’t enough (multi-replica coherence, persistence), layer a remote store under the memory cache — cache-aside, not a different API. Implement the documented CacheStore shape (in the package README) against your backend; the README carries the full recipe and a getUser walkthrough.