@lifeomic/gradual-feature-toggle-tools

gradual-feature-toggle-tools

A set of tools to make gradual rollout of features easier

Selecting users to get a feature

When a new feature is being rolled out and is currently 50% enabled, then ideally 50% of the users would get the new behavior. The enablement of the new behavior should be uniform as the feature value is increased and should not be biased by the usernames or feature names. For example, if you use e-mail addresses then ideally the enablement of @gmail.com users would be evenly spread across the range of enablement. The testcases assert that uniform distribution is true.

Given a feature an a user, this code can help you decide whether a user should receive the new behavior or not:

import { calculateUserEnablementThreshold } from 'gradual-feature-toggle-tools';

const featureValue = // some lookup of a feature's current value (from 0 - 100)
const userThreshold = await calculateUserEnablementThreshold('some-feature', 'some-user');
if (featureValue > userThredshold) {
    // Use the new behavior
} else {
    // Use the old behavior
}