Swift Confidence SDK

This repo contains the official Swift SDK for accessing feature flags and for event tracking with Confidence.

It also contains the Confidence OpenFeature Provider, to be used in conjunction with the OpenFeature SDK.

For documentation related to flags management and event tracking in Confidence, refer to the Confidence documentation website.

Functionalities:

• Managed integration with the Confidence backend
• Prefetch and cache flag evaluations, for fast value reads even when the application is offline
• Automatic data collection about which flags have been accessed by the application
• Event tracking for instrumenting your application

Using Confidence with OpenFeature

We suggest that you use Confidence together with the OpenFeature SDK. This means that your app interacts completely with the OpenFeature SDK for feature flagging, and the Confidence Provider will be the engine for realizing the feature flagging values.

Swift Package Manager

In the dependencies section of Package.swift add:

.package(url: "[email protected]:spotify/confidence-sdk-swift.git", from: "1.4.4")

and in the target dependencies section add:

.product(name: "Confidence", package: "confidence-sdk-swift"),
.product(name: "ConfidenceProvider", package: "confidence-sdk-swift"),

Create and set the Provider

The Confidence Provider instance needs to be created and then set in the global OpenFeatureAPI. The Confidence Provider takes in the configured Confidence instance for its initialization:

import Confidence
import ConfidenceProvider
import OpenFeature

let confidence = Confidence.Builder(clientSecret: "mysecret", loggerLevel: .NONE).build()
let provider = ConfidenceFeatureProvider(confidence: confidence)
let ctx = ImmutableContext(targetingKey: "myTargetingKey", structure: ImmutableStructure())
OpenFeatureAPI.shared.setProvider(provider: provider, initialContext: ctx)

The evaluation context is the way for the client to specify contextual data that Confidence uses to evaluate rules defined on the flag.

The setProvider() function is synchronous and returns immediately, however this does not mean that the provider is ready to be used. An asynchronous network request to the Confidence backend to fetch all the flags configured for your application must be completed by the provider first. The provider will then emit a READY event indicating you can start resolving flags.

There is also an async/await compatible API available for waiting the Provider to become ready:

await OpenFeatureAPI.shared.setProviderAndWait(provider: provider)

A utility function is available on the provider to check if the current storage has any stored values - this can be used to determine the best initialization strategy.

// If we have no cache, then do a fetch first.
var initializationStrategy: InitializationStrategy = .activateAndFetchAsync
if ConfidenceFeatureProvider.isStorageEmpty() {
    initializationStrategy = .fetchAndActivate
}

Initialization strategies:

• activateAndFetchAsync: the flags in the cached are used for this session, while updated values are fetched and stored on disk for a future session; this means that a READY event is immediately emitted when calling setProvider();
• fetchAndActivate: the Provider attempts to refresh the flag cache on disk before exposing the flags; this might prolong the time needed for the Provider to become READY.

To listen for the READY event, you can add an event handler via the OpenFeatureAPI shared instance:

OpenFeatureAPI.shared.observe().sink { event in
    if event == .ready {
        // Provider is ready
    }
}

Note: if you do attempt to resolve a flag before the READY event is emitted, you may receive the default value with the reason STALE.

There are other events that are emitted by the provider, see Provider Events in the Open Feature specification for more details.

Updating the Evaluation Context

It is possible to update the evaluation context within an application's session via the following API:

let ctx = ImmutableContext(targetingKey: "myNewTargetingKey", structure: ImmutableStructure())
OpenFeatureAPI.shared.setEvaluationContext(evaluationContext: ctx)

setEvaluationContext() is a synchronous function similar to setProvider(). It calls the Confidence backend to fetch the flag evaluations according to the new evaluation context; if the call is successful, it replaces the cache with the new flag data.

Notes:

• The initialization strategy is not taken into consideration when calling setEvaluationContext(), so it's required to wait for READY before resuming to resolve flags.

• If you do attempt to resolve a flag before the READY event is emitted, you may receive the old value with the reason STALE.

• A "targeting key" in the evaluation context is expected by the OpenFeature APIs, but a different custom field inside the structure value can also be configured as the randomization unit in the Confidence portal. In this case, it's okay to leave targetingKey empty.

Handling Provider Errors

When calling setEvaluationContext() or setProvider() via the OpenFeatureAPI an ERROR event can be emitted if something goes wrong.

To listen for the ERROR event, you can add an event handler via the OpenFeatureAPI shared instance:

OpenFeatureAPI.shared.observe().sink { event in
    if event == .error {
        // An error has been emitted
    }
}

Request a flag / value

The client is used to retrieve values for the current user / context. For example, retrieving a boolean value for the flag my-flag.my-boolean:

let client = OpenFeatureAPI.shared.getClient()
let result = client.getBooleanValue(key: "my-flag.my-boolean", defaultValue: false)

In Confidence each flag value is a complex data structure including one or more properties of different types. To access a specific flag's property, the dot notation in the example above is used. The full data structure for a flag can always be fetched via:

let result = client.getObjectValue(key: "my-flag", defaultValue: Value.null)

Note: if a flag can't be resolved from the local cache, the provider doesn't automatically resort to calling remote. Refreshing the cache from remote only happens when setting a new provider and/or evaluation context in the global OpenFeatureAPI.

Confidence Vanilla SDK

If you want to use Confidence without OpenFeature, you can. Please take a look at our dedicated SDK readme.