@graphile/lru

You probably want lru-cache instead.

This is an obsessively optimized LRU cache for Node.js, it forgoes features in favour of performance, and is very marginally faster than node-lru-cache v6 in certain circumstances (namely those that we care about inside the Graphile internals). A performance comparison versus node-lru-cache v7 has not yet been performed.

Usage

import LRU from "@graphile/lru";

// Construct an LRU cache
const lru = new LRU({
  // Maximum number of items to store (required)
  maxLength: 500, // Must be an integer at least 2

  // Optional callback to handle when a value is removed (both when LRU becomes
  // full and when LRU is reset)
  dispose(key, value) {
    console.log(`Disposing of key '${key}' with value '${inspect(value)}'`);
  },
});

const ANY_KEY_HERE = { foo: "bar" };
const ANY_VALUE_HERE = { randomNumber: () => 4 };
// Store a value
lru.set(ANY_KEY_HERE, ANY_VALUE_HERE);

// Retrieve a value
const value = lru.get(ANY_KEY_HERE);

// Clear the cache
lru.reset();

Considering a pull request?

Pull requests that add features that aren't required by the Graphile suite are unlikely to be entertained - use node-lru-cache instead!

Pull requests that improve performance should come with benchmarks, benchmark scripts and explanation of technique. And if you pull it off without breaking anything - YES PLEASE.

Pull requests that add documentation are welcome.