46 KiB
title |
---|
useMemo |
useMemo
is a React Hook that lets you cache the result of a calculation between re-renders.
const memoizedValue = useMemo(calculateValue, dependencies)
Usage {/usage/}
Skipping expensive recalculations {/skipping-expensive-recalculations/}
By default, React will re-run the entire body of your component every time that it re-renders. For example, if this TodoList
updates its state or receives new props from its parent, the filterTodos
function will re-run:
function TodoList({ todos, tab, theme }) {
const visibleTodos = filterTodos(todos, tab);
// ...
}
Usually, this isn't a problem because most calculations are very fast. However, if you're filtering or transforming a large array, or doing some expensive computation, you might want to skip doing it again if data hasn't changed. If both todos
and tab
are the same as they were during the last render, you can instruct React to reuse the visibleTodos
you've already calculated during the last render. This type of caching is called memoization.
To cache a value between re-renders, wrap its calculation in a useMemo
call at the top level of your component:
import { useMemo } from 'react';
function TodoList({ todos, tab, theme }) {
const visibleTodos = useMemo(() => filterTodos(todos, tab), [todos, tab]);
// ...
}
You need to pass two things to useMemo
:
- A calculation function that takes no arguments, like
() =>
, and returns what you wanted to calculate. - A list of dependencies including every value within your component that's used inside your calculation.
On the initial render, the value you'll get from useMemo
will be the result of calling your calculation.
On every subsequent render, React will compare the dependencies with the dependencies you passed during the last render. If none of the dependencies have changed (compared with Object.is
), useMemo
will return the value you already calculated on the last render. Otherwise, React will re-run your calculation and return the new value.
In other words, useMemo
will cache your function's result, and return it on re-renders until the dependencies change. If both todos
and tab
are the same as before, the TodoList
won't have to recalculate visibleTodos
.
You should only rely on useMemo
as a performance optimization. If your code doesn't work without it, find the underlying problem and fix it first. Then you may add useMemo
to improve performance.
In general, unless you're creating or looping over thousands of objects, it's probably not expensive. If you want to get more confidence, you can add a console log to measure the time spent in a piece of code:
console.time('filter array');
const visibleTodos = filterTodos(todos, tab);
console.timeEnd('filter array');
Perform the interaction you're measuring (for example, typing into the input). You will then see logs like filter array: 0.15ms
in your console. If the overall logged time adds up to a significant amount (say, 1ms
or more), it might make sense to memoize that calculation. As an experiment, you can then wrap the calculation in useMemo
to verify whether the total logged time has decreased for that interaction or not:
console.time('filter array');
const visibleTodos = useMemo(() => {
return filterTodos(todos, tab); // Skipped if todos and tab haven't changed
}, [todos, tab]);
console.timeEnd('filter array');
useMemo
won't make the first render faster. It only helps you skip unnecessary work on updates.
Keep in mind that your machine is probably faster than your users' so it's a good idea to test the performance with an artificial slowdown. For example, Chrome offers a CPU Throttling option for this.
Also note that measuring performance in development will not give you the most accurate results. (For example, when Strict Mode is on, you will see each component render twice rather than once.) To get the most accurate timings, build your app for production and test it on a device like your users have.
If your app is like this site, and most interactions are coarse (like replacing a page or an entire section), memoization is usually unnecessary. On the other hand, if your app is more like a drawing editor, and most interactions are granular (like moving shapes), then you might find memoization very helpful.
Optimizing with useMemo
is only valuable in a few cases:
- The calculation you're putting in
useMemo
is noticeably slow, and its dependencies rarely change. - You pass it as a prop to a component wrapped in
memo
. You want to skip re-rendering if the value hasn't changed. Memoization lets your component re-render only when dependencies are the same. - The value you're passing is later used as a dependency of some Hook. For example, maybe another
useMemo
calculation value depends on it. Or maybe you are depending on this value fromuseEffect.
There is no benefit to wrapping a calculation in useMemo
in other cases. There is no significant harm to doing that either, so some teams choose to not think about individual cases, and memoize as much as possible. The downside of this approach is that code becomes less readable. Also, not all memoization is effective: a single value that's "always new" is enough to break memoization for an entire component.
In practice, you can make a lot of memoization unnecessary by following a few principles:
- When a component visually wraps other components, let it accept JSX as children. This way, when the wrapper component updates its own state, React knows that its children don't need to re-render.
- Prefer local state and don't lift state up any further than necessary. For example, don't keep transient state like forms and whether an item is hovered at the top of your tree or in a global state library.
- Keep your rendering logic pure. If re-rendering a component causes a problem or produces some noticeable visual artifact, it's a bug in your component! Fix the bug instead of adding memoization.
- Avoid unnecessary Effects that update state. Most performance problems in React apps are caused by chains of updates originating from Effects that cause your components to render over and over.
- Try to remove unnecessary dependencies from your Effects. For example, instead of memoization, it's often simpler to move some object or a function inside an Effect or outside the component.
If a specific interaction still feels laggy, use the React Developer Tools profiler to see which components would benefit the most from memoization, and add memoization where needed. These principles make your components easier to debug and understand, so it's good to follow them in any case. In the long term, we're researching doing granular memoization automatically to solve this once and for all.
Skipping recalculation with useMemo
{/skipping-recalculation-with-usememo/}
In this example, the filterTodos
implementation is artificially slowed down so that you can see what happens when some JavaScript function you're calling during rendering is genuinely slow. Try switching the tabs and toggling the theme.
Switching the tabs feels slow because it forces the slowed down filterTodos
to re-execute. That's expected because the tab
has changed, and so the entire calculation needs to re-run. (If you're curious why it runs twice, it's explained here.)
Next, try toggling the theme. Thanks to useMemo
, it's fast despite the artificial slowdown! The slow filterTodos
call was skipped because both todos
and tab
(which you pass as dependencies to useMemo
) haven't changed since the last render.
import { useState } from 'react';
import { createTodos } from './utils.js';
import TodoList from './TodoList.js';
const todos = createTodos();
export default function App() {
const [tab, setTab] = useState('all');
const [isDark, setIsDark] = useState(false);
return (
<>
<button onClick={() => setTab('all')}>
All
</button>
<button onClick={() => setTab('active')}>
Active
</button>
<button onClick={() => setTab('completed')}>
Completed
</button>
<br />
<label>
<input
type="checkbox"
checked={isDark}
onChange={e => setIsDark(e.target.checked)}
/>
Dark mode
</label>
<hr />
<TodoList
todos={todos}
tab={tab}
theme={isDark ? 'dark' : 'light'}
/>
</>
);
}
import { useMemo } from 'react';
import { filterTodos } from './utils.js'
export default function TodoList({ todos, theme, tab }) {
const visibleTodos = useMemo(
() => filterTodos(todos, tab),
[todos, tab]
);
return (
<div className={theme}>
<p><b>Note: <code>filterTodos</code> is artificially slowed down!</b></p>
<ul>
{visibleTodos.map(todo => (
<li key={todo.id}>
{todo.completed ?
<s>{todo.text}</s> :
todo.text
}
</li>
))}
</ul>
</div>
);
}
export function createTodos() {
const todos = [];
for (let i = 0; i < 50; i++) {
todos.push({
id: i,
text: "Todo " + (i + 1),
completed: Math.random() > 0.5
});
}
return todos;
}
export function filterTodos(todos, tab) {
console.log('[ARTIFICIALLY SLOW] Filtering ' + todos.length + ' todos for "' + tab + '" tab.');
let startTime = performance.now();
while (performance.now() - startTime < 500) {
// Do nothing for 500 ms to emulate extremely slow code
}
return todos.filter(todo => {
if (tab === 'all') {
return true;
} else if (tab === 'active') {
return !todo.completed;
} else if (tab === 'completed') {
return todo.completed;
}
});
}
label {
display: block;
margin-top: 10px;
}
.dark {
background-color: black;
color: white;
}
.light {
background-color: white;
color: black;
}
Always recalculating a value {/always-recalculating-a-value/}
In this example, the filterTodos
implementation is also artificially slowed down so that you can see what happens when some JavaScript function you're calling during rendering is genuinely slow. Try switching the tabs and toggling the theme.
Unlike in the previous example, toggling the theme is also slow now! This is because there is no useMemo
call in this version, so the artificially slowed down filterTodos
gets called on every re-render. It is called even if only theme
has changed.
import { useState } from 'react';
import { createTodos } from './utils.js';
import TodoList from './TodoList.js';
const todos = createTodos();
export default function App() {
const [tab, setTab] = useState('all');
const [isDark, setIsDark] = useState(false);
return (
<>
<button onClick={() => setTab('all')}>
All
</button>
<button onClick={() => setTab('active')}>
Active
</button>
<button onClick={() => setTab('completed')}>
Completed
</button>
<br />
<label>
<input
type="checkbox"
checked={isDark}
onChange={e => setIsDark(e.target.checked)}
/>
Dark mode
</label>
<hr />
<TodoList
todos={todos}
tab={tab}
theme={isDark ? 'dark' : 'light'}
/>
</>
);
}
import { filterTodos } from './utils.js'
export default function TodoList({ todos, theme, tab }) {
const visibleTodos = filterTodos(todos, tab);
return (
<div className={theme}>
<ul>
<p><b>Note: <code>filterTodos</code> is artificially slowed down!</b></p>
{visibleTodos.map(todo => (
<li key={todo.id}>
{todo.completed ?
<s>{todo.text}</s> :
todo.text
}
</li>
))}
</ul>
</div>
);
}
export function createTodos() {
const todos = [];
for (let i = 0; i < 50; i++) {
todos.push({
id: i,
text: "Todo " + (i + 1),
completed: Math.random() > 0.5
});
}
return todos;
}
export function filterTodos(todos, tab) {
console.log('[ARTIFICIALLY SLOW] Filtering ' + todos.length + ' todos for "' + tab + '" tab.');
let startTime = performance.now();
while (performance.now() - startTime < 500) {
// Do nothing for 500 ms to emulate extremely slow code
}
return todos.filter(todo => {
if (tab === 'all') {
return true;
} else if (tab === 'active') {
return !todo.completed;
} else if (tab === 'completed') {
return todo.completed;
}
});
}
label {
display: block;
margin-top: 10px;
}
.dark {
background-color: black;
color: white;
}
.light {
background-color: white;
color: black;
}
However, here is the same code with the artificial slowdown removed. Does the lack of useMemo
feel noticeable or not?
import { useState } from 'react';
import { createTodos } from './utils.js';
import TodoList from './TodoList.js';
const todos = createTodos();
export default function App() {
const [tab, setTab] = useState('all');
const [isDark, setIsDark] = useState(false);
return (
<>
<button onClick={() => setTab('all')}>
All
</button>
<button onClick={() => setTab('active')}>
Active
</button>
<button onClick={() => setTab('completed')}>
Completed
</button>
<br />
<label>
<input
type="checkbox"
checked={isDark}
onChange={e => setIsDark(e.target.checked)}
/>
Dark mode
</label>
<hr />
<TodoList
todos={todos}
tab={tab}
theme={isDark ? 'dark' : 'light'}
/>
</>
);
}
import { filterTodos } from './utils.js'
export default function TodoList({ todos, theme, tab }) {
const visibleTodos = filterTodos(todos, tab);
return (
<div className={theme}>
<ul>
{visibleTodos.map(todo => (
<li key={todo.id}>
{todo.completed ?
<s>{todo.text}</s> :
todo.text
}
</li>
))}
</ul>
</div>
);
}
export function createTodos() {
const todos = [];
for (let i = 0; i < 50; i++) {
todos.push({
id: i,
text: "Todo " + (i + 1),
completed: Math.random() > 0.5
});
}
return todos;
}
export function filterTodos(todos, tab) {
console.log('Filtering ' + todos.length + ' todos for "' + tab + '" tab.');
return todos.filter(todo => {
if (tab === 'all') {
return true;
} else if (tab === 'active') {
return !todo.completed;
} else if (tab === 'completed') {
return todo.completed;
}
});
}
label {
display: block;
margin-top: 10px;
}
.dark {
background-color: black;
color: white;
}
.light {
background-color: white;
color: black;
}
Quite often, code without memoization works fine. If your interactions are fast enough, you might not need memoization.
You can try increasing the number of todo items in utils.js
and see how the behavior changes. This particular calculation wasn't very expensive to begin with, but if the number of todos grows significantly, most of the overhead will be in re-rendering rather than in the filtering. Keep reading below to see how you can optimize re-rendering with useMemo
.
Skipping re-rendering of components {/skipping-re-rendering-of-components/}
By default, when a component re-renders, React re-renders all of its children recursively. This is fine for components that don't require much calculation to re-render. Components higher up the tree or slower components can opt into skipping re-renders when their props are the same by wrapping themselves in memo
:
import { memo } from 'react';
function List({ items }) {
// ...
}
export default memo(List);
This is a performance optimization. The useMemo
and useCallback
Hooks are often needed to make it work.
For this optimization to work, the parent component that renders this <List />
needs to ensure that, if it doesn't want List
to re-render, every prop it passes to the List
must be the same as on the last render.
Let's say the parent TodoList
component looks like this:
export default function TodoList({ todos, tab, theme }) {
const visibleTodos = filterTodos(todos, tab);
return (
<div className={theme}>
<List items={visibleTodos} />
</div>
);
}
With the above code, the List
optimization will not work because visibleTodos
will be a different array on every re-render of the TodoList
component. To fix it, wrap the calculation of visibleTodos
in useMemo
:
export default function TodoList({ todos, tab, theme }) {
const visibleTodos = useMemo(() => filterTodos(todos, tab), [todos, tab]);
return (
<div className={theme}>
<List items={visibleTodos} />
</div>
);
}
After this change, as long as todos
and tab
haven't changed, thanks to useMemo
, the visibleTodos
won't change between re-renders. Since List
is wrapped in memo
, it will only re-render if one of its props is different from its value on the last render. You're passing the same items
prop, so List
can skip the re-rendering entirely.
Notice that in this example, it doesn't matter whether filterTodos
itself is fast or slow. The point isn't to avoid a slow calculation, but it's to avoid passing a different prop value every time since that would break the memo
optimization of the child List
component. The useMemo
call in the parent makes memo
work for the child.
Instead of wrapping List
in memo
, you could wrap the <List />
JSX node itself in useMemo
:
export default function TodoList({ todos, tab, theme }) {
const visibleTodos = useMemo(() => filterTodos(todos, tab), [todos, tab]);
const children = useMemo(() => <List items={visibleTodos} />, [visibleTodos]);
return (
<div className={theme}>
{children}
</div>
);
}
The behavior would be the same. If the visibleTodos
haven't changed, List
won't be re-rendered.
A JSX node like <List items={visibleTodos} />
is an object like { type: List, props: { items: visibleTodos } }
. Creating this object is very cheap, but React doesn't know whether its contents is the same as last time or not. This is why by default, React will re-render the List
component.
However, if React sees the same exact JSX as during the previous render, it won't try to re-render your component. This is because JSX nodes are immutable. A JSX node object could not have changed over time, so React knows it's safe to skip a re-render. However, for this to work, the node has to actually be the same object, not merely look the same in code. This is what useMemo
does in this example.
Manually wrapping JSX nodes into useMemo
is not convenient. For example, you can't do this conditionally. This is usually you would wrap components with memo
instead of wrapping JSX nodes.
Skipping re-rendering with useMemo
and memo
{/skipping-re-rendering-with-usememo-and-memo/}
In this example, the List
component is artificially slowed down so that you can see what happens when a React component you're rendering is genuinely slow. Try switching the tabs and toggling the theme.
Switching the tabs feels slow because it forces the slowed down List
to re-render. That's expected because the tab
has changed, and so you need to reflect the user's new choice on the screen.
Next, try toggling the theme. Thanks to useMemo
together with memo
, it’s fast despite the artificial slowdown! The List
skipped re-rendering because the visibleItems
array has not changed since the last render. The visibleItems
array has not changed because both todos
and tab
(which you pass as dependencies to useMemo
) haven't changed since the last render.
import { useState } from 'react';
import { createTodos } from './utils.js';
import TodoList from './TodoList.js';
const todos = createTodos();
export default function App() {
const [tab, setTab] = useState('all');
const [isDark, setIsDark] = useState(false);
return (
<>
<button onClick={() => setTab('all')}>
All
</button>
<button onClick={() => setTab('active')}>
Active
</button>
<button onClick={() => setTab('completed')}>
Completed
</button>
<br />
<label>
<input
type="checkbox"
checked={isDark}
onChange={e => setIsDark(e.target.checked)}
/>
Dark mode
</label>
<hr />
<TodoList
todos={todos}
tab={tab}
theme={isDark ? 'dark' : 'light'}
/>
</>
);
}
import { useMemo } from 'react';
import List from './List.js';
import { filterTodos } from './utils.js'
export default function TodoList({ todos, theme, tab }) {
const visibleTodos = useMemo(
() => filterTodos(todos, tab),
[todos, tab]
);
return (
<div className={theme}>
<p><b>Note: <code>List</code> is artificially slowed down!</b></p>
<List items={visibleTodos} />
</div>
);
}
import { memo } from 'react';
function List({ items }) {
console.log('[ARTIFICIALLY SLOW] Rendering <List /> with ' + items.length + ' items');
let startTime = performance.now();
while (performance.now() - startTime < 500) {
// Do nothing for 500 ms to emulate extremely slow code
}
return (
<ul>
{items.map(item => (
<li key={item.id}>
{item.completed ?
<s>{item.text}</s> :
item.text
}
</li>
))}
</ul>
);
}
export default memo(List);
export function createTodos() {
const todos = [];
for (let i = 0; i < 50; i++) {
todos.push({
id: i,
text: "Todo " + (i + 1),
completed: Math.random() > 0.5
});
}
return todos;
}
export function filterTodos(todos, tab) {
return todos.filter(todo => {
if (tab === 'all') {
return true;
} else if (tab === 'active') {
return !todo.completed;
} else if (tab === 'completed') {
return todo.completed;
}
});
}
label {
display: block;
margin-top: 10px;
}
.dark {
background-color: black;
color: white;
}
.light {
background-color: white;
color: black;
}
Always re-rendering a component {/always-re-rendering-a-component/}
In this example, the List
implementation is also artificially slowed down so that you can see what happens when some React component you're rendering is genuinely slow. Try switching the tabs and toggling the theme.
Unlike in the previous example, toggling the theme is also slow now! This is because there is no useMemo
call in this version, so the visibleTodos
is always a different array, and the slowed down List
component can't skip re-rendering.
import { useState } from 'react';
import { createTodos } from './utils.js';
import TodoList from './TodoList.js';
const todos = createTodos();
export default function App() {
const [tab, setTab] = useState('all');
const [isDark, setIsDark] = useState(false);
return (
<>
<button onClick={() => setTab('all')}>
All
</button>
<button onClick={() => setTab('active')}>
Active
</button>
<button onClick={() => setTab('completed')}>
Completed
</button>
<br />
<label>
<input
type="checkbox"
checked={isDark}
onChange={e => setIsDark(e.target.checked)}
/>
Dark mode
</label>
<hr />
<TodoList
todos={todos}
tab={tab}
theme={isDark ? 'dark' : 'light'}
/>
</>
);
}
import List from './List.js';
import { filterTodos } from './utils.js'
export default function TodoList({ todos, theme, tab }) {
const visibleTodos = filterTodos(todos, tab);
return (
<div className={theme}>
<p><b>Note: <code>List</code> is artificially slowed down!</b></p>
<List items={visibleTodos} />
</div>
);
}
import { memo } from 'react';
function List({ items }) {
console.log('[ARTIFICIALLY SLOW] Rendering <List /> with ' + items.length + ' items');
let startTime = performance.now();
while (performance.now() - startTime < 500) {
// Do nothing for 500 ms to emulate extremely slow code
}
return (
<ul>
{items.map(item => (
<li key={item.id}>
{item.completed ?
<s>{item.text}</s> :
item.text
}
</li>
))}
</ul>
);
}
export default memo(List);
export function createTodos() {
const todos = [];
for (let i = 0; i < 50; i++) {
todos.push({
id: i,
text: "Todo " + (i + 1),
completed: Math.random() > 0.5
});
}
return todos;
}
export function filterTodos(todos, tab) {
return todos.filter(todo => {
if (tab === 'all') {
return true;
} else if (tab === 'active') {
return !todo.completed;
} else if (tab === 'completed') {
return todo.completed;
}
});
}
label {
display: block;
margin-top: 10px;
}
.dark {
background-color: black;
color: white;
}
.light {
background-color: white;
color: black;
}
However, here is the same code with the artificial slowdown removed. Does the lack of useMemo
feel noticeable or not?
import { useState } from 'react';
import { createTodos } from './utils.js';
import TodoList from './TodoList.js';
const todos = createTodos();
export default function App() {
const [tab, setTab] = useState('all');
const [isDark, setIsDark] = useState(false);
return (
<>
<button onClick={() => setTab('all')}>
All
</button>
<button onClick={() => setTab('active')}>
Active
</button>
<button onClick={() => setTab('completed')}>
Completed
</button>
<br />
<label>
<input
type="checkbox"
checked={isDark}
onChange={e => setIsDark(e.target.checked)}
/>
Dark mode
</label>
<hr />
<TodoList
todos={todos}
tab={tab}
theme={isDark ? 'dark' : 'light'}
/>
</>
);
}
import List from './List.js';
import { filterTodos } from './utils.js'
export default function TodoList({ todos, theme, tab }) {
const visibleTodos = filterTodos(todos, tab);
return (
<div className={theme}>
<List items={visibleTodos} />
</div>
);
}
import { memo } from 'react';
function List({ items }) {
return (
<ul>
{items.map(item => (
<li key={item.id}>
{item.completed ?
<s>{item.text}</s> :
item.text
}
</li>
))}
</ul>
);
}
export default memo(List);
export function createTodos() {
const todos = [];
for (let i = 0; i < 50; i++) {
todos.push({
id: i,
text: "Todo " + (i + 1),
completed: Math.random() > 0.5
});
}
return todos;
}
export function filterTodos(todos, tab) {
return todos.filter(todo => {
if (tab === 'all') {
return true;
} else if (tab === 'active') {
return !todo.completed;
} else if (tab === 'completed') {
return todo.completed;
}
});
}
label {
display: block;
margin-top: 10px;
}
.dark {
background-color: black;
color: white;
}
.light {
background-color: white;
color: black;
}
Quite often, code without memoization works fine. If your interactions are fast enough, you don't need memoization.
Keep in mind that you need to run React in production mode, disable React Developer Tools, and use devices similar to the ones your app's users have in order to get a realistic sense of what's actually slowing down your app.
Memoizing a dependency of another Hook {/memoizing-a-dependency-of-another-hook/}
Suppose you have a calculation that depends on an object created directly in the component body:
function Dropdown({ allItems, text }) {
// This object is created directly in the component body
const searchOptions = { matchMode: 'whole-word', text };
const visibleItems = useMemo(() => {
return searchItems(allItems, searchOptions);
}, [allItems, searchOptions]); // 🚩 Caution: Dependency on an object created in the component body
// ...
Depending on an object like this defeats the point of memoization. When a component re-renders, all of the code directly inside the component body runs again. The lines of code creating the searchOptions
object will also run on every re-render. Since searchOptions
is a dependency of your useMemo
call, and it's different every time, React will know the dependencies are different from the last time, and recalculate searchItems
every time.
To fix this, you could memoize the searchOptions
object itself before passing it as a dependency:
function Dropdown({ allItems, text }) {
const searchOptions = useMemo(() => {
return { matchMode: 'whole-word', text };
}, [text]); // ✅ Only changes when text changes
const visibleItems = useMemo(() => {
return searchItems(allItems, searchOptions);
}, [allItems, searchOptions]); // ✅ Only changes when allItems or searchOptions changes
// ...
In the example above, if the text
did not change, the searchOptions
object also won't change. However, an even better fix is to move the searchOptions
object declaration inside of the useMemo
calculation function:
function Dropdown({ allItems, text }) {
const visibleItems = useMemo(() => {
// ✅ This object is created inside useMemo
const searchOptions = { matchMode: 'whole-word', text };
return searchItems(allItems, searchOptions);
}, [allItems, text]); // ✅ Only changes when allItems or text changes
// ...
Now your calculation depends on text
directly (which is a string and can't "accidentally" be new like an object).
You can use a similar approach to prevent useEffect
from firing again unnecessarily. Before you try to optimize dependencies with useMemo
, see if you can make them unnecessary. Read about removing Effect dependencies.
Memoizing a function {/memoizing-a-function/}
Suppose the Form
component is wrapped in memo
. You want to pass a function to it as a prop:
export default function ProductPage({ productId, referrer }) {
function handleSubmit(orderDetails) {
post('/product/' + productId + '/buy', {
referrer,
orderDetails
});
}
return <Form onSubmit={handleSubmit} />;
}
Similar to how {}
always creates a different object, function declarations like function() {}
and expressions like () => {}
produce a different function on every re-render. By itself, creating a new function is not a problem. This is not something to avoid! However, if the Form
component is memoized, presumably you want to skip re-rendering it when no props have changed. A prop that is always different would defeat the point of memoization.
To memoize a function with useMemo
, your calculation function would have to return another function:
export default function Page({ productId, referrer }) {
const handleSubmit = useMemo(() => {
return (orderDetails) => {
post('/product/' + product.id + '/buy', {
referrer,
orderDetails
});
};
}, [productId, referrer]);
return <Form onSubmit={handleSubmit} />;
}
This looks clunky! Memoizing functions is common enough that React has a built-in Hook specifically for that. Wrap your functions into useCallback
instead of useMemo
to avoid having to write an extra nested function:
export default function Page({ productId, referrer }) {
const handleSubmit = useCallback((orderDetails) => {
post('/product/' + product.id + '/buy', {
referrer,
orderDetails
});
}, [productId, referrer]);
return <Form onSubmit={handleSubmit} />;
}
The two examples above are completely equivalent. The only benefit to useCallback
is that it lets you avoid writing an extra nested function inside. It doesn't do anything else. Read more about useCallback
.
Reference {/reference/}
useMemo(calculateValue, dependencies)
{/usememo/}
Call useMemo
at the top level of your component to declare a memoized value:
import { useMemo } from 'react';
function TodoList({ todos, tab }) {
const visibleTodos = useMemo(
() => filterTodos(todos, tab),
[todos, tab]
);
// ...
}
Parameters {/parameters/}
-
calculateValue
: The function calculating the value that you want to memoize. It should be pure, should take no arguments, and should return a value of any type. React will call your function during the initial render. On subsequent renders, React will return the same value again if thedependencies
have not changed since the last render. Otherwise, it will callcalculateValue
, return its result, and store it in case it can be reused later. -
dependencies
: The list of all reactive values referenced inside of thecalculateValue
code. Reactive values include props, state, and all the variables and functions declared directly inside your component body. If your linter is configured for React, it will verify that every reactive value is correctly specified as a dependency. The list of dependencies must have a constant number of items and be written inline like[dep1, dep2, dep3]
. React will compare each dependency with its previous value using theObject.is
comparison algorithm.
Returns {/returns/}
On the initial render, useMemo
returns the result of calling calculateValue
with no arguments.
During subsequent renders, it will either return an already stored value from the last render (if the dependencies haven't changed), or call calculateValue
again, and return the result that calculateValue
has returned.
Caveats {/caveats/}
useMemo
is a Hook, so you can only call it at the top level of your component or your own Hooks. You can't call it inside loops or conditions. If you need that, extract a new component and move the state into it.- In Strict Mode, React will call your calculation function twice in order to help you find accidental impurities. This is development-only behavior and does not affect production. If your calculation function is pure (as it should be), this should not affect the logic of your component. The result from one of the calls will be ignored.
- React will not throw away the cached value unless there is a specific reason to do that. For example, in development, React throws away the cache when you edit the file of your component. Both in development and in production, React will throw away the cache if your component suspends during the initial mount. In the future, React may add more features that take advantage of throwing away the cache--for example, if React adds built-in support for virtualized lists in the future, it would make sense to throw away the cache for items that scroll out of the virtualized table viewport. This should match your expectations if you rely on
useMemo
solely as a performance optimization. Otherwise, a state variable or a ref may be more appropriate.
Troubleshooting {/troubleshooting/}
My calculation runs twice on every re-render {/my-calculation-runs-twice-on-every-re-render/}
In Strict Mode, React will call some of your functions twice instead of once:
function TodoList({ todos, tab }) {
// This component function will run twice for every render.
const visibleTodos = useMemo(() => {
// This calculation will run twice if any of the dependencies change.
return filterTodos(todos, tab);
}, [todos, tab]);
// ...
This is expected and shouldn't break your code.
This development-only behavior helps you keep components pure. React uses the result of one of the calls, and ignores the result of the other call. As long as your component and calculation functions are pure, this shouldn't affect your logic. However, if they are accidentally impure, this helps you notice the mistakes and fix it.
For example, this impure calculation function mutates an array you received as a prop:
const visibleTodos = useMemo(() => {
// 🚩 Mistake: mutating a prop
todos.push({ id: 'last', text: 'Go for a walk!' });
const filtered = filterTodos(todos, tab);
return filtered;
}, [todos, tab]);
Because React calls your calculation twice, you'll see the todo was added twice, so you'll know that there is a mistake. Your calculation can't change the objects that it received, but it can change any new objects you created during the calculation. For example, if filterTodos
always returns a different array, you can mutate that array:
const visibleTodos = useMemo(() => {
const filtered = filterTodos(todos, tab);
// ✅ Correct: mutating an object you created during the calculation
filtered.push({ id: 'last', text: 'Go for a walk!' });
return filtered;
}, [todos, tab]);
Read keeping components pure to learn more about purity.
Also, check out the guides on updating objects and updating arrays without mutation.
My useMemo
call is supposed to return an object, but returns undefined {/my-usememo-call-is-supposed-to-return-an-object-but-returns-undefined/}
This code doesn't work:
// 🔴 You can't return an object from an arrow function with () => {
const searchOptions = useMemo(() => {
matchMode: 'whole-word',
text: text
}, [text]);
In JavaScript, () => {
starts the arrow function body, so the {
brace is not a part of your object. This is why it doesn't return an object, and leads to confusing mistakes. You could fix it by adding parentheses like ({
and })
:
// This works, but is easy for someone to break again
const searchOptions = useMemo(() => ({
matchMode: 'whole-word',
text: text
}), [text]);
However, this is still confusing and too easy for someone to break by removing the parentheses.
To avoid this mistake, write a return
statement explicitly:
// ✅ This works and is explicit
const searchOptions = useMemo(() => {
return {
matchMode: 'whole-word',
text: text
};
}, [text]);
Every time my component renders, the calculation in useMemo
re-runs {/every-time-my-component-renders-the-calculation-in-usememo-re-runs/}
Make sure you've specified the dependency array as a second argument!
If you forget the dependency array, useMemo
will re-run the calculation every time:
function TodoList({ todos, tab }) {
// 🔴 Recalculates every time: no dependency array
const visibleTodos = useMemo(() => filterTodos(todos, tab));
// ...
This is the corrected version passing the dependency array as a second argument:
function TodoList({ todos, tab }) {
// ✅ Does not recalculate unnecessarily
const visibleTodos = useMemo(() => filterTodos(todos, tab), [todos, tab]);
// ...
If this doesn't help, then the problem is that at least one of your dependencies is different from the previous render. You can debug this problem by manually logging your dependencies to the console:
const visibleTodos = useMemo(() => filterTodos(todos, tab), [todos, tab]);
console.log([todos, tab]);
You can then right-click on the arrays from different re-renders in the console and select "Store as a global variable" for both of them. Assuming the first one got saved as temp1
and the second one got saved as temp2
, you can then use the browser console to check whether each dependency in both arrays is the same:
Object.is(temp1[0], temp2[0]); // Is the first dependency the same between the arrays?
Object.is(temp1[1], temp2[1]); // Is the second dependency the same between the arrays?
Object.is(temp1[2], temp2[2]); // ... and so on for every dependency ...
When you find which dependency is breaking memoization, either find a way to remove it, or memoize it as well.
All my component's props are memoized, but it still re-renders every time {/all-my-components-props-are-memoized-but-it-still-re-renders-every-time/}
There are three possible reasons for this:
- Your component (or some Hook it uses) updates its state, but a re-render wasn't necessary.
- Your component is reading context, and that context has updated, but a re-render wasn't necessary.
- Your component accepts
children
as a prop, so it always receives different JSX.
To solve the first two problems, split your component into two: an outer one, and a memoized inner one.
This lets you add memoization in the middle between them without changing any of the parent components:
export default function FormWrapper(props) {
const { formSettings } = useSettings();
return <Form {...props} formSettings={formSettings} />
}
function Form(props) {
// ...
}
Form = memo(Form);
If FormWrapper
re-renders but formSettings
haven't changed, it will immediately skip re-rendering Form
.
Now let's see how to recognize and solve the last problem (a component accepting JSX re-renders every time). Imagine this FancyBorder
component is wrapped in memo
. However, it re-renders even if theme
doesn't change:
function TodoList({ todos, tab }) {
const visibleTodos = useMemo(() => filterTodos(todos, tab), [todos, tab]);
return (
<FancyBorder theme={theme}>
<List items={visibleTodos} />
</FancyBorder>
);
}
This is because it accepts a piece of JSX as the children
prop:
function TodoList({ todos, tab }) {
const visibleTodos = useMemo(() => filterTodos(todos, tab), [todos, tab]);
return (
<FancyBorder theme={theme}>
<List items={visibleTodos} />
</FancyBorder>
);
}
A JSX node like <List items={visibleTodos} />
produces an object like { type: List, props: { items: visibleTodos } }
. Creating this object is very cheap, but React doesn't know whether its contents is the same as last time or not. This is why by default, React will re-render the List
component. If you need to prevent FancyBorder
from re-rendering when todos
or tab
change, you could memoize its JSX node itself:
function TodoList({ todos, tab }) {
const visibleTodos = useMemo(() => filterTodos(todos, tab), [todos, tab]);
const children = useMemo(() => <List items={visibleTodos} />, [visibleTodos]);
return (
<FancyBorder theme={theme}>
{children}
</FancyBorder>
);
}
Alternatively, to prevent FancyBorder
from re-rendering when the todos change, move it up the tree above the component that holds the todo items in state. Then React would not need to re-render it on most interactions:
function App({ theme }) {
return (
<Layout>
<Sidebar />
<FancyBorder theme={theme}>
<MainContent />
</FancyBorder>
<Footer />
</Layout>
);
}
I need to call useMemo
for each list item in a loop, but it's not allowed {/i-need-to-call-usememo-for-each-list-item-in-a-loop-but-its-not-allowed/}
You can't call useMemo
in a loop:
function ReportList({ items }) {
return (
<article>
{items.map(item => {
// 🔴 You can't call useMemo in a loop like this:
const data = useMemo(() => calculateReport(item), [item]);
return (
<figure key={data.id}>
<Chart data={data} />
</figure>
);
})}
</article>
);
}
Instead, extract a component for each item and memoize data for individual items:
function ReportList({ items }) {
return (
<article>
{items.map(item =>
<Report key={item.id} item={item} />
)}
</article>
);
}
function Report({ item }) {
// ✅ Call useMemo at the top level:
const data = useMemo(() => calculateReport(item), [item]);
return (
<figure>
<Chart data={data} />
</figure>
);
}
Report = memo(Report); // ✅ Memoize individual items