dflag v3.0.0-5

Flag

Feature flagging made easy for React and Redux

yarn add flag

Motivation

Feature flagging is necessary for large client-side applications. They improve development speed and allow teams to test new features before they are stable. In order to WANT to use feature flags in an application, they should be VERY easy to add and remove. That means minimal boiler plate and no need to pass boolean props down through component hierarchy. Such a thing could be done with global variables; however, they live outside of the React/Redux lifecycle, making them more difficult to control. Instead, this library injects and then accesses feature flags directly from the React context without getting in your way.

Getting Started

Flag allows you to declare flags as either plain values or as functions. If a flag is a function then it is referred to as a computed flag. The function accepts one argument which is the flags object itself. You do not have to use computed flags, but they can be very convenient. For example:

const flags = {
  // properties can be nested objects
  features: {
    // they can be boolean
    useMyCoolNewThing: true,
  },
  config: {
    // they can be strings
    apiUrl: 'www.example.com/api',
  },
  // they can be numbers
  cool: 1,
  dude: 5,
  // they can be computed
  coolAndDude: flags => flags.cool + flags.dude,
  // they can be computed from other computed properties.
  // other computed properties are resolved for you, so that you do not
  // need to call it as a function.
  largeCoolAndDude: flags => flags.coolAndDude > 10
};

Notice that computed flags nested in other flags (e.g. flags.coolAndDude inside largeCoolAndDude) is already resolved as its value. The flags object can then either be store in your Redux store or passed in as a prop to FlagProvider.

General use

This library can be used with vanilla React and with React-Redux. The main component is Flag that specifies which flag you're checking and how to handle it. Use this component whenever you need to split rendering based on a flag.

The decision trees of which component or function to call is done in the following order:

• Is the flag truthy?
  • Yes
    • Does the component have a child(ren)?
      • Yes
        • Render the child(ren)
        • DONE.
      • No
        • Does the component have a component prop?
          • Yes
            • Render an instance of that component
            • DONE.
          • No
            • Does the component have a render prop?
              • Yes
                • Call the function and use the return.
                • DONE.
              • No
                • Render null
                • DONE.
  • No
    • Does the component have a fallbackComponent prop?
      • Yes
        • Render an instance of that component
        • DONE.
      • No
        • Does the component have a fallbackRender prop?
          • Yes
            • Call the function and use the return.
            • DONE.
          • No
            • Render null
            • DONE.

Here's an example of using render and fallbackRender, forking on features.useMyCoolNewThing.

import { Flag } from 'flag';

<Flag
  name="features.useMyCoolNewThing"
  render={() =>
    <div>Rendered on truthy</div>
  }
  fallbackRender={() =>
    <div>Rendered on falsy</div>
  }
/>

Given that we have full components to be rendered in each path, we could also use component and fallbackComponent.

import { Flag } from 'flag';

<Flag
  name="features.useMyCoolNewThing"
  component={RevisedFeature}
  fallbackComponent={ExistingFeature}
/>

If you don't care about the fallback case - render nothing if false - then you can also render your component inline as children.

import { Flag } from 'flag';

<Flag name="features.useMyCoolNewThing">
  <RevisedFeature />
</Flag>

Use with react

To use this with just React, we handle flags with the FlagProvider component which makes flags available to child through React context.

import { FlagsProvider, Flag } from 'flag';

const flags = { /*...*/ };

const instance = (
  <FlagsProvider flags={flags}>
    <div>
      This is my application.
      <Flag
        name="features.useMyCoolNewThing"
        render={() =>
          <div>Rendered on truthy</div>
        }
        fallbackRender={() =>
          <div>Rendered on falsy</div>
        }
      />
    </div>
  </FlagsProvider>
);

React.render(instance, document.querySelector('#app'));

Use with react-redux

You can alternatively keep your flags in a Redux store. The only caveat here is that they must be store on the flags key of your state. In the example below, we use createFlagsReducer to create the correct reducer.

import { createStore, combineReducers } from 'redux';
import { createFlagsReducer } from 'flag';

const reducer = combineReducer({
  ...myOtherReducers,
  flags: createFlagsReducer({
    // properties can be nested objects
    features: {
      // they can be boolean
      useMyCoolNewThing: true,
    },
    config: {
      // they can be strings
      apiUrl: 'www.example.com/api',
    },
    // they can be numbers
    cool: 1,
    dude: 5,
    // they can be computed
    coolAndDude: flags => flags.cool + flags.dude,
    // they can be computed from other computed properties.
    largeCoolAndDude: flags => flags.coolAndDude > 10
  })
});

const store = createStore(reducer);

After creating the store, we attach flags to the correct context by wrapping the application in ConnectedFlagsProvider which retrieves the flag state. Then the Flag component behaves as usual.

import { Provider } from 'react-redux';
import { ConnectedFlagsProvider, Flag } from 'flag';

const instance = (
  <Provider store={store}>
    <ConnectedFlagsProvider>
      <div>
        This is my application.
        <Flag
          name="features.useMyCoolNewThing"
          render={() =>
            <div>Rendered on truthy</div>
          }
          fallbackRender={() =>
            <div>Rendered on falsy</div>
          }
        />
      </div>
    </ConnectedFlagsProvider>
  </Provider>
);

React.render(instance, document.querySelector('#app'));

API

Flag

The main React component.

<Flag
  name="flagA"
  render={(valueOfFlagA) => <TruthyFeature />}
  fallbackRender={(valueOfFlagA) => <FalsyFeature />}
/>

FlagsProvider

Attaches flags to the appropriate React context. Also transforms computed flags.

<FlagsProvider flags={{myFeature: true}}>
  <App />
</FlagsProvider>

ConnectedFlagsProvider

Same as FlagsProvider except flags are fetched from a Redux store which has been attached to React context by the React-Redux Provider.

<Provider store={store}>
  <ConnectedFlagsProvider>
    <App />
  </ConnectedFlagsProvider>
</Provider>

setFlagsAction

A dispatchable action that sets flags.

store.dispatch(
  setFlagsAction({
    myFeature: false
  })
);

createFlagsReducer

Creates the reducer for your Redux store. Accepts any plain object as an argument.

const myDefaultFlags = {
  features: {
    useMyCoolNewThing: true,
    useMyOtherThing: false,
    proAccount: ({features}) => features.useMyCoolNewThing && features.useMyOtherThing
  },
  config: {
    apiUrl: 'www.example.com/api',
  },
}

const reducer = combineReducers({
  ...myOtherReducers
  flags: createFlagsReducer(myDefaultFlags)
})

License

MIT