Video: Feature Flagging with React Server Components

Hey, everybody, I'm Graham, and I'm really excited to give you a talk today on feature flagging with React server components. We're going to give examples of how you can deploy feature flags in your application, and how feature flagging interacts with React's rendering strategies. Feature flags are conditional logic that allows you to control the state of a feature independent from the code deploy. This separation of code deployments from feature releases is critical for large product organizations. It enables frequent commits without exposing changes to users, conditional release of features, and easy rollback in case of unexpected bugs. Using FeatureFlag gives you a kill switch where you can turn off your feature. It allows you to do A-B testing and measure the impact of new features. You conditionally release the feature to different sets of users, allowing you to control for externalities and changes to your product. A FeatureFlag is used to get the state and conditionally show a component. The flag state can be stored in a file or using environment variables. However, environment variables have limitations in terms of complexity. A more complex example uses growth book syntax, allowing for advanced control over feature rollout. Building your own system can be tricky, but feature flagging platforms offer fully-featured solutions with two main advantages. Feature flagging applications have a nice UI for controlling feature releases and an SDK that integrates with the code. React poses complications in using feature flags due to its rendering strategies. Static site generators like Next.js have limitations in user targeting and require redeploy for updates. Client components are not async, requiring the use of React primitives. These approaches have their own challenges in implementing feature flags. To initialize the SDK, instantiate the SDK outside of the app and use a user effect hook to download the feature flag payload and update the user data on the SDK. The growth book provider builds context for evaluating feature flags, allowing components to easily access feature flag states. However, there may be a slight flicker in the UX due to the time it takes to initialize the SDK and download the feature flag state. Network optimization can help reduce flickering, but the client's network connection is beyond your control. There are some workarounds for flickering, like showing a loading spinner while rendering in the background. Using feature flags for SEO is not ideal as the initial HTML payload doesn't include all the contents. Server components in React 19 provide an async solution without the need for complicated useEffect and state. react.cache allows caching expensive operations scoped to the current request. The getSdk function call retrieves the value of the feature flag from the cache, providing faster subsequent calls. Our SDK has its own in-memory cache. Dynamic rendering can be expensive in terms of requests, rendering, network calls, and running React on the service side. It can also be slow, and tracking events may be challenging. The server-client hybrid approach combines the advantages of the previous strategies without the downsides. By making the outer app a server component and caching the feature flag payload, we can achieve no-flicker client-side feature flagging. The client component can use the UseMemo hook and the initSync method with the payload already in memory. Passing the SDK instance and wrapping the main app in a provider allows us to have the best of both worlds. Although it may be slightly more complex to set up, using Next.js with React server components offers a cool and modern approach to feature flagging with high performance.