In computing, memoization or memoisation
is an optimization technique used primarily
to speed up computer programs by storing
the results of expensive function calls and Â
returning the cached result when the same
inputs occur again. Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â
                                                    — wikipedia
It's important to memoize heavy computations as well as arrays and object creations so that they don't get re-created on every render. A re-render occurs when state changes, redux dispatches some action, or when the user types into a text input (re-render for every single key press). You don't want to run a lot of operations in those renders for very obvious reasons - so no heavy filtering, no list operations, etc.
A Pure Component (or a React.memo
component) does not re-render if it's props are shallow equal.
Each variable you create in your render function will get re-allocated on each render. While this is not a problem for value types, this causes reference types to be different on every render. When you pass those variables down to pure components via props, they will still re-render even though the props are logically the same. Often those variables even go over the Bridge and make your app slow.
When a pure component re-renders, it compares the previous props to the current props and checks if they are shallow-equal.
Numbers, strings and booleans are value types, which means they can be compared by value:
const i1 = 7;
const i2 = 7;
const equal = i1 === i2; // true
Objects, arrays and functions are reference types, which means they cannot be compared by their logical value, but have to be compared by reference:
const o1 = { x: 7 };
const o2 = { x: 7 };
const equal = o1 === o2; // false
Reference comparisons simply compare the memory address of the variable, so only o1 === o1
would be true
in the above code example.
There are libraries like deep-equal to compare objects by actual equality, but that's not shallow equality anymore.
If you create objects in your render function, they will be re-created on every single render. This means when you create an object in the first render, it is not reference-equal to the object in the second render. For this very reason, memoization exists.
- Use the
useMemo
hook to memoize arrays and objects which will keep their reference equality (and won't get re-created on each render) as long as the dependencies (second argument) stay the same. Also useuseMemo
to cache heavy computations, such as array operations, filtering, etc. - Use the
useCallback
hook to memoize a function.
In general, function components can be optimized more easily due to the concept of hooks. You can however apply similar techniques for class components, just be aware that this will result in a lot more code.
While animations and performance intensive tasks are scheduled on native threads, your entire business logic runs on a single JavaScript thread, so make sure you're doing as little work as possible there. Doing too much work on the JavaScript thread can be compared to a high ping in a video game - you can still look around smoothly, but you can't really play the game because every interaction takes too long.
Native components (<View>
, <Text>
, <Image>
, <Blurhash>
, ...) have to pass props to native via the bridge. They can be memoized, so React compares the props for shallow-equality and only passes them over the bridge if they are different than the props from the last render. If you don't memoize correctly, you might up passing props over the bridge for every single render, causing the bridge to be very occupied. See the Styles example - styles will get sent over the bridge on every re-render!
Here are a few examples to help you avoid doing too much work on your JavaScript thread:
return <View style={[styles.container, { backgroundColor: 'red' }]} />
const style = useStyle(() => [styles.container, { backgroundColor: 'red' }], []);
return <View style={style} />
- Reanimated styles from
useAnimatedStyle
, as those have to be dynamic.
See react-native-style-utilities for the
useStyle
hook
Using filter
, map
or other array operations in renderers will run the entire operation again for every render.
return <Text>{users.filter((u) => u.status === "online").length} users online</Text>
const onlineCount = useMemo(() => users.filter((u) => u.status === "online").length, [users]);
return <Text>{onlineCount} users online</Text>
You can also apply this to render multiple React views with .map
. Those can be memoized with useMemo
too.
return <View onLayout={(layout) => console.log(layout)} />
const onLayout = useCallback((layout) => {
console.log(layout);
}, []);
return <View onLayout={onLayout} />
Make sure to also think about other calls in the renderer, e.g. useSelector
, useComponentDidAppear
- wrap the callback there too!
function MyComponent(props) {
return <PressableOpacity onPress={() => props.logoutUser()} />
}
function MyComponent(props) {
return <PressableOpacity onPress={props.logoutUser} />
}
function MyComponent(props) {
return <RecyclerListView scrollViewProps={{ horizontal: props.isHorizontal }} />;
}
function MyComponent(props) {
const scrollViewProps = useMemo(() => ({
horizontal: props.isHorizontal
}), [props.isHorizontal]);
return <RecyclerListView scrollViewProps={scrollViewProps} />;
}
function MyComponent() {
return <RecyclerListView scrollViewProps={{ horizontal: true }} />;
}
const SCROLL_VIEW_PROPS = { horizontal: true }
function MyComponent() {
return <RecyclerListView scrollViewProps={SCROLL_VIEW_PROPS} />;
}
This applies to objects as well as functions which don't depend on the component's state or props. Always use this if you can, since it's even more efficient than useMemo
and useCallback
.
const [me, setMe] = useState(users.find((u) => u.id === myUserId));
const [me, setMe] = useState(() => users.find((u) => u.id === myUserId));
The useState
hook accepts an initializer function. While the first example ("Bad") runs the .find
on every render, the second example only runs the passed function once to initialize the state.
When writing new components I always put a log statement in my render function to passively watch how often my component re-renders while I'm working on it. In general, components should re-render as little as possible, and if I see a lot of logs appearing in my console I know I did something wrong. It's a good practice to put this function in your component once you start working on it, and remove it once done.
function ComponentImWorkingOn() {
// code
console.log('re-rendering ComponentImWorkingOn!');
return <View />;
}
You can also use the why-did-you-render library to find out why a component has re-rendered (prop changes, state changes, ...) and possibly catch mistakes early on.
export const MyComponent = (props) => {
return ...
}
const MyComponentImpl = (props) => {
return ...
}
export const MyComponent = React.memo(MyComponentImpl);
If your component renders the same result given the same props, you can wrap it in a call to React.memo(...)
for a performance boost in some cases by memoizing the result. This means that React will skip rendering the component, and reuse the last rendered result. See the official docs for React.memo
, and use React.memo(...)
wisely.
If your app feels slow, try the react-native-performance library and it's flipper plugin to profile your app's performance in various aspects such as time to interactive, component render time, script execution and more.
memoize!!