Configuration

configureClient(...) wires the client key, HTTP, and the anon-id store. Call it once at app launch. It returns the ShipeasyClient and registers it as the process-global one, fetchable with shipeasyClient(). It is first-config-wins (idempotent) — the first call configures; later calls return the same client and change nothing. See installation for where to place the call.

import Shipeasy

let client = configureClient(clientKey: "pk_live_…")

On first config it fire-and-forgets an anonymous identify([:]), so getFlag resolves for logged-out users without an explicit identify.

configureClient() options

Environment-derived egress defaults

By default the SDK is quiet outside production: on a dev machine or in CI it makes no outbound request unless you opt in. Both isNetworkEnabled and isTrackingEnabled default to on in production and off in every other environment, so an app embedding the SDK never phones home from a local build.

"Production" is decided in this order:

1. The native env var SHIPEASY_ENV, then APP_ENV, then ENV. A value of production or prod (case-insensitive) means production; any other present value (development, staging, test, …) means not production.
2. If none is set (the common case for a shipped app), the #if DEBUG build flag: a debug build is not production, a release build is.
3. If still undecided, the env: option you pass (which itself defaults to "prod"), so a real production deploy stays on.

An explicitly-passed value always overrides the environment default:

// Force the client fully offline (e.g. a preview build), even in production:
configureClient(clientKey: "pk_live_…", isNetworkEnabled: false)

// Force it on in a non-production build (e.g. a staging QA app):
configureClient(clientKey: "pk_live_…", isNetworkEnabled: true)

Behaviour change (2.1.0): earlier versions always made network calls. To restore that everywhere, either set SHIPEASY_ENV=production in the app's environment or pass isNetworkEnabled: true.

The configured client — shipeasyClient()

configureClient(...) returns the client and stores it globally. Anywhere in the app, fetch it with shipeasyClient() (returns ShipeasyClient?, or nil before configuration):

let flag = await shipeasyClient()?.getFlag("new_checkout") ?? false

Binding the user — identify

identify(_:) binds the device user and refreshes assignments over POST /sdk/evaluate. Pass your targeting attribute map; the persisted anonymous_id is always attached automatically:

await shipeasyClient()?.identify(["user_id": "u_123", "plan": "pro"])

Call identify:

• at launch (or [:] for a logged-out visitor — the anonymous identify on first config already covers this if you don't need attributes),
• on login, with user_id and any targeting attributes,
• whenever targeting attributes change (plan upgrade, opted-in setting, etc.).

Awaiting identify guarantees the first reads see the fresh assignments. A failed evaluate is non-fatal — reads simply fall back to the supplied defaults.

Logout — reset

reset() clears the bound user_id but keeps the device anonymous_id, then re-evaluates as an anonymous visitor:

await shipeasyClient()?.reset()

Re-evaluate — refreshAssignments

refreshAssignments() re-evaluates for the current user without changing identity — e.g. to pick up a just-published flag:

await shipeasyClient()?.refreshAssignments()

The persisted anonymous_id

The stable device id is exposed as await shipeasyClient()?.anonymousId. It is minted once and persisted via the store so it survives cold starts. This is what keeps a logged-out visitor in the same bucket for every fractional rollout and experiment on every launch. See advanced to back it with the Keychain or an app-group container.