Evaluation

The client offers evaluating single feature flags based on feature flag names or evaluating all feature flags at once.

Please also read about the Evaluation Context and the Emergency mode.

Single feature flag evaluation

When calling one of the "single feature flag" evaluation functions, the client ensures to return a value with the correct data type.

Boolean

To evaluate a feature flag that is defined as a boolean feature flag in the Loli specification, use evaluateBooleanFeatureFlag. You need to pass the name of the feature flag and the evaluation context data. The function returns a promise/is async that resolves with the evaluation result.

const evaluationContext : EvaluationContext = { /*...*/ }

const booleanValue = await client.evaluateBooleanFeatureFlag(
    "boolean-feature-flag-name",
    evaluationContext
);

Number

To evaluate a feature flag that is defined as a number feature flag in the Loli specification, use evaluateNumberFeatureFlag. You need to pass the name of the feature flag and the evaluation context data. The function returns a promise/is async that resolves with the evaluation result.

const evaluationContext : EvaluationContext = { /*...*/ }

const numberValue = await client.evaluateNumberFeatureFlag(
    "number-feature-flag-name",
    evaluationContext
);

String

To evaluate a feature flag that is defined as a string feature flag in the Loli specification, use evaluateStringFeatureFlag. You need to pass the name of the feature flag and the evaluation context data. The function returns a promise/is async that resolves with the evaluation result.

const evaluationContext : EvaluationContext = { /*...*/ }

const stringValue = await client.evaluateStringFeatureFlag(
    "string-feature-flag-name",
    evaluationContext
);

All feature flags evaluation

If you want to evaluate all feature flags, use evaluateAllFeatureFlags. You only need to pass the evaluation context data. The function returns a promise/is async that resolves with an object mapping feature flag names to feature flag values.

const evaluationContext : EvaluationContext = { /*...*/ }

const allFeatureFlagValues = await client.evaluateAllFeatureFlags(
    evaluationContext
);

console.log("Boolean feature flag =", allFeatureFlagValues['boolean-feature-flag-name']);
console.log("Number feature flag =", allFeatureFlagValues['number-feature-flag-name']);
console.log("String feature flag =", allFeatureFlagValues['string-feature-flag-name']);

Evaluation context

The "evaluation context" is the data the SDK uses to evaluate property conditions that are used by feature flags and segments.

It has to be an object that can contain anything – but ideally carries all the data that has been defined in the Loli specification as "properties".

Data can be structured in a flat object:

const evaluationContext : EvaluationContext = {
    id: "8as65d876as5d",
    email: "test@test.com",
    appearance: "dark-mode",
    notifactions: "all"
}

But you can also pass a nested object or use arrays:

const evaluationContext : EvaluationContext = {
    id: "8as65d876as5d",
    email: "test@test.com",
    settings: {
        appearance: "dark-mode",
        notifactions: "all",
        activatedFeatures: ["ai-pilot", "8bit-mode"]
    }
}

Emergency mode

The client can act in an "emergency mode" and feature flag evaluations will return emergency fallback values.

These are defined by via the client's options.

The client returns emergency fallback values in the following cases:

• The spec loader never succeeded.
• The spec loader did not return a valid Loli specification.
• The client could not find a feature flag definition in the loaded specification for a given feature flag name (single feature flag evaluation).
• The data type of a feature flag defined in the specification various from the evaluation function signature (single feature flag evaluation).

Segment Caching

The Loli specification supports segment specification which act as reusable sets of conditions. By nature, segments are ideally used by multiple feature flags or event other segments. This prevents defining the same conditions in e.g. multiple feature flags over and over again.

To not evaluate one segment multiple times during a single feature flag or all feature flags evaluation operation, a LoliClient uses a segment evaluation cache.

Whenever a segment is evaluated during an evaluation call it will not be reevaluated again during the same evaluation call. Instead, the cached evaluation is then used.

This works, because all internal evaluations are deterministic. Same evaluation context. Same evaluation output.

TIP

Make use of segments as much as possible to speed up the feature flag evaluation for very large feature flag specifications.