Caching strategy

Start by matching each kind of data to one of three storage tiers:

1. 1.Memory (NSCache, in-process stores): fastest access, and gone when the process dies. Use it for decoded images and the session's data models.
2. 2.Disk (files, URLCache): survives restarts, so it suits large blobs and structured data, and a schema version in each key retires old entries after a migration. Keep each response's ETag (entity tag) so the next request can send If-None-Match and get 304 Not Modified instead of the full body.
3. 3.Database (Core Data, SwiftData, or a SQLite wrapper such as GRDB): best when you need queries, relationships, or incremental updates rather than key-value blobs. Treat it as the source of truth while offline, with a clear rule for when a row counts as stale.

let memoryCapacity = 50 * 1_024 * 1_024 // 50 MB in RAM
let diskCapacity = 200 * 1_024 * 1_024 // 200 MB on disk
let cacheDir = FileManager.default.urls(for: .cachesDirectory, in: .userDomainMask).first

let cache = URLCache(
    memoryCapacity: memoryCapacity,
    diskCapacity: diskCapacity,
    directory: cacheDir
)

// Attach the cache to a session configuration
let config = URLSessionConfiguration.default
config.urlCache = cache

// Cache first: serve any cached copy regardless of age, with no revalidation.
// The default .useProtocolCachePolicy honors Cache-Control and ETag headers instead.
config.requestCachePolicy = .returnCacheDataElseLoad

let session = URLSession(configuration: config)

Cache resources that are read often and need to show immediately, plus content the user explicitly saves or revisits. Don't cache sensitive data without encryption, and keep secrets in the Keychain rather than in any cache.