shouldComponentUpdate in function components

I have a question regarding React's shouldComponentUpdate (when not overwritten). I do prefer pure, function components, but I am afraid that it updates every time, even though props/state did not change. So I am considering using the PureComponent class instead.

My question regarding that: Do function components have the same shouldComponentUpdate check as PureComponents? Or does it update every time?

A functional component will rerender every time the parent renders it, no matter if the props have changed or not.

However, using the React.memo high order component it is actually possible for functional components to get the same shouldComponentUpdate check which is used in PureComponent https://reactjs.org/docs/react-api.html#reactmemo

You can simply wrap your functional component in React.memo on export like shown here.

So

const SomeComponent = (props) => (<div>HI</div>)
export default SomeComponent

Could instead be

const SomeComponent = (props) => (<div>HI</div>)
export default React.memo(SomeComponent)

Example

The following example show how this affects rerenders

The parent component is just a regular functional component. It is using the new react hooks to handle some state updates.

It just have some tick state which only serves the purpose of giving some clues on how often we rerender a prop, while it forces a rerender of the parent component twice a second.

Further we have a clicks state which tell how often we have clicked the button. This is what the prop we send to the children. Hence they should ONLY rerender if the number of clicks changes if we use React.memo

Now notice that we have two different kind of children. One wrapped in memo and one which is not. Child which is not wrapped, will rerender every time the parent rerenders. MemoChild which is wrapped, will only rerender when the clicks property changes.

const Parent = ( props ) => {
  // Ticks is just some state we update 2 times every second to force a parent rerender
  const [ticks, setTicks] = React.useState(0);
  setTimeout(() => setTicks(ticks + 1), 500);
  // The ref allow us to pass down the updated tick without changing the prop (and forcing a rerender)
  const tickRef = React.useRef();
  tickRef.current = ticks;

  // This is the prop children are interested in
  const [clicks, setClicks] = React.useState(0);

  return (
    <div>
      <h2>Parent Rendered at tick {tickRef.current} with clicks {clicks}.</h2>
      <button 
        onClick={() => setClicks(clicks + 1)}>
        Add extra click
      </button>
      <Child tickRef={tickRef} clicks={clicks}/>
      <MemoChild tickRef={tickRef} clicks={clicks}/>
    </div>
  );
};

const Child = ({ tickRef, clicks }) => (
  <p>Child Rendered at tick {tickRef.current} with clicks {clicks}.</p>
);

const MemoChild = React.memo(Child);

Live Example (also on CodePen):

console.log("HI");

const Parent = ( props ) => {
  const [ticks, setTicks] = React.useState(0);
  const tickRef = React.useRef();
  tickRef.current = ticks;
  const [clicks, setClicks] = React.useState(0);
  
  setTimeout(() => setTicks(ticks + 1), 500);
  return (
    <div>
      <h2>Parent Rendered at tick {tickRef.current} with clicks {clicks}.</h2>
      <button 
        onClick={() => setClicks(clicks + 1)}>
        Add extra click
      </button>
      <Child tickRef={tickRef} clicks={clicks}/>
      <MemoChild tickRef={tickRef} clicks={clicks}/>
    </div>
  );
};

const Child = ({ tickRef, clicks }) => (
  <p>Child Rendered at tick {tickRef.current} with clicks {clicks}.</p>
);

const MemoChild = React.memo(Child);


ReactDOM.render(
  <Parent />,
  document.getElementById('root')
);

<div id="root"></div>
<script src="https://cdnjs.cloudflare.com/ajax/libs/react/16.12.0/umd/react.production.min.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/react-dom/16.12.0/umd/react-dom.production.min.js"></script>

In React, functional components are stateless and they do not have lifecycle methods. Stateless components are an elegant way of writing React components without much code in our bundle. But internally, Stateless components are wrapped in a class without any optimizations currently applied. That means both stateless and stateful components has the same code path internally (although we define them differently).

But in the future React may optimize stateless components as said it here:

In the future, we’ll also be able to make performance optimizations specific to these components by avoiding unnecessary checks and memory allocations. [ More reading... ]

shouldComponentUpdate

This is where you we can apply our custom optimizations and avoid unnecessary re-rendering of components. The usage of this method with different types of components are explained below:

• Functional Stateless components

  As said before, stateless components do not have life cycle methods thus we cannot optimize them using shouldComponentUpdate. But they are already optimized in a different way, they have much simpler and elegant code structure and costs less bytes than a component with all life-cycle hooks.

• extend React.PureComponent

  From React v15.3.0, we have a new base class called PureComponent to extend with PureRenderMixin built-in. Under the hood this employs a shallow comparison of current props/state with next props/state within a shouldComponentUpdate.

  That said, we still cannot rely on PureComponent class to get our components optimized to the level we want. This anomaly case happens if we have props with Object types (arrays, dates, plain objects). This is because we have this problem when comparing objects:

  const obj1 = { id: 1 };
  const obj2 = { id: 1 };
  console.log(obj1 === obj2); // prints false
  

  Hence a shallow comparison won't suffice to determine whether things have changed or not. But use PureComponent class if your props are just string, number, boolean.. and not objects. Also use it if you do not want to implement your own custom optimizations.

• extend React.Component

  Consider the above example; if we know that the objects have changed if the id has changed, then we can implement our own custom optimization by comparing obj1.id === obj2.id. That is where we can extend our normal Component base class and use shouldComponentUpdate to do the comparison of specific keys by ourselves.

Another approach is to use useMemo to update values only when watched ones are updated:

const memoizedValue = useMemo(() => computeExpensiveValue(a, b), [a, b]);

In case of objects there's an option of using state hook to cache the value of interested variable, after making sure it's updated. Eg by using lodash:

const [fooCached, setFooCached]: any = useState(null);

if (!_.isEqual(fooCached, foo)) {
  setFooCached(foo);
}). 

const showFoo = useMemo(() => {
    return <div>Foo name: { foo.name }</div>
}, [fooCached]);