---
title: API
type: reference
summary: Complete API reference for SWR hooks and utilities.
prerequisites:
- /docs/getting-started
related:
- /docs/global-configuration
- /docs/typescript
---
# API
```js
const { data, error, isLoading, isValidating, mutate } = useSWR(key, fetcher, options)
```
## Parameters
* `key`: a unique key string for the request (or a function / array / null) [(details)](/docs/arguments), [(advanced usage)](/docs/conditional-fetching)
* `fetcher`: (*optional*) a Promise-returning function to fetch your data [(details)](/docs/data-fetching)
* `options`: (*optional*) an object of options for this SWR hook
## Return Values
* `data`: data for the given key resolved by `fetcher` (or undefined if not loaded)
* `error`: error thrown by `fetcher` (or undefined)
* `isLoading`: if there's an ongoing request and no "loaded data". Fallback data and previous data are not considered "loaded data"
* `isValidating`: if there's a request or revalidation loading
* `mutate(data?, options?)`: function to mutate the cached data [(details)](/docs/mutation)
More information can be found [here](/docs/advanced/understanding).
## Options
* `suspense = false`: enable React Suspense mode [(details)](/docs/suspense)
* `fetcher(args)`: the fetcher function
* `revalidateIfStale = true`: automatically revalidate even if there is stale data [(details)](/docs/revalidation#disable-automatic-revalidations)
* `revalidateOnMount`: enable or disable automatic revalidation when component is mounted [(details)](/docs/revalidation#revalidate-on-mount)
* `revalidateOnFocus = true`: automatically revalidate when window gets focused [(details)](/docs/revalidation)
* `revalidateOnReconnect = true`: automatically revalidate when the browser regains a network connection (via `navigator.onLine`) [(details)](/docs/revalidation)
* `refreshInterval` [(details)](/docs/revalidation):
* Disabled by default: `refreshInterval = 0`
* If set to a number, polling interval in milliseconds
* If set to a function, the function will receive the latest data and should return the interval in milliseconds
* `refreshWhenHidden = false`: polling when the window is invisible (if `refreshInterval` is enabled)
* `refreshWhenOffline = false`: polling when the browser is offline (determined by `navigator.onLine`)
* `shouldRetryOnError = true`: retry when fetcher has an error
* `dedupingInterval = 2000`: dedupe requests with the same key in this time span in milliseconds
* `focusThrottleInterval = 5000`: only revalidate once during a time span in milliseconds
* `loadingTimeout = 3000`: timeout to trigger the onLoadingSlow event in milliseconds
* `errorRetryInterval = 5000`: error retry interval in milliseconds
* `errorRetryCount`: max error retry count
* `fallback`: a key-value object of multiple fallback data [(example)](/docs/with-nextjs)
* `fallbackData`: initial data to be returned (note: This is per-hook)
* `strictServerPrefetchWarning = false`: show a warning message in the console when a key has no pre-filled data provided [(details)](/docs/with-nextjs#prefetch-data-in-server-components)
* `keepPreviousData = false`: return the previous key's data until the new data has been loaded [(details)](/docs/advanced/understanding#return-previous-data-for-better-ux)
* `onLoadingSlow(key, config)`: callback function when a request takes too long to load (see `loadingTimeout`)
* `onSuccess(data, key, config)`: callback function when a request finishes successfully
* `onError(err, key, config)`: callback function when a request returns an error
* `onErrorRetry(err, key, config, revalidate, revalidateOps)`: handler for error retry
* `onDiscarded(key)`: callback function when a request is ignored due to race conditions
* `compare(a, b)`: comparison function used to detect when returned data has changed, to avoid spurious rerenders. By default, [stable-hash](https://github.com/shuding/stable-hash) is used.
* `isPaused()`: function to detect whether pause revalidations, will ignore fetched data and errors when it returns `true`. Returns `false` by default.
* `use`: array of middleware functions [(details)](/docs/middleware)
When under a slow network (2G, {'<='} 70Kbps), errorRetryInterval will be 10s, and{' '}
loadingTimeout will be 5s by default.
You can also use [global configuration](/docs/global-configuration) to provide default options.
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Arguments
type: reference
summary: Pass arguments to SWR fetcher functions.
prerequisites:
- /docs/getting-started
related:
- /docs/data-fetching
- /docs/conditional-fetching
---
# Arguments
By default, `key` will be passed to `fetcher` as the argument. So the following 3 expressions are equivalent:
```js
useSWR('/api/user', () => fetcher('/api/user'))
useSWR('/api/user', url => fetcher(url))
useSWR('/api/user', fetcher)
```
## Multiple Arguments
In some scenarios, it's useful to pass multiple arguments (can be any value or object) to the `fetcher` function.
For example an authorized fetch request:
```js
useSWR('/api/user', url => fetchWithToken(url, token))
```
This is **incorrect**. Because the identifier (also the cache key) of the data is `'/api/user'`,
even if `token` changes, SWR will still use the same key and return the wrong data.
Instead, you can use an **array** as the `key` parameter, which contains multiple arguments of `fetcher`:
```js
const { data: user } = useSWR(['/api/user', token], ([url, token]) => fetchWithToken(url, token))
```
The `fetcher` function accepts the `key` parameter as is, and the cache key will also be associated with the entire `key` argument. In the above example, `url` and `token` are both tied to the cache key.
In the previous versions (\< 2.0.0), The `fetcher` function will receive the spreaded arguments from original `key` when the `key` argument is array type. E.g., key `[url, token]` will become 2 arguments `(url, token)` for `fetcher` function.
## Passing Objects
Since SWR 1.1.0, object-like keys will be serialized under the hood automatically.
Say you have another function that fetches data with a user scope: `fetchWithUser(api, user)`. You can do the following:
```js
const { data: user } = useSWR(['/api/user', token], fetchWithToken)
// ...and then pass it as an argument to another useSWR hook
const { data: orders } = useSWR(user ? ['/api/orders', user] : null, fetchWithUser)
```
You can directly pass an object as the key, and `fetcher` will receive that object too:
```js
const { data: orders } = useSWR({ url: '/api/orders', args: user }, fetcher)
```
In older versions (\< 1.1.0), SWR **shallowly** compares the arguments on every render, and triggers revalidation if any of them has changed.
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Conditional Fetching
type: guide
summary: Conditionally fetch data based on dependencies or user state.
prerequisites:
- /docs/data-fetching
related:
- /docs/data-fetching
- /docs/arguments
---
# Conditional Fetching
## Conditional
Use `null` or pass a function as `key` to conditionally fetch data. If the function throws or returns a falsy value, SWR will not start the request.
```js
// conditionally fetch
const { data } = useSWR(shouldFetch ? '/api/data' : null, fetcher)
// ...or return a falsy value
const { data } = useSWR(() => shouldFetch ? '/api/data' : null, fetcher)
// ...or throw an error when user.id is not defined
const { data } = useSWR(() => '/api/data?uid=' + user.id, fetcher)
```
## Dependent
SWR also allows you to fetch data that depends on other data. It ensures the maximum possible parallelism (avoiding waterfalls), as well as serial fetching when a piece of dynamic data is required for the next data fetch to happen.
```js
function MyProjects () {
const { data: user } = useSWR('/api/user')
const { data: projects } = useSWR(() => '/api/projects?uid=' + user.id)
// When passing a function, SWR will use the return
// value as `key`. If the function throws or returns
// falsy, SWR will know that some dependencies are not
// ready. In this case `user.id` throws when `user`
// isn't loaded.
if (!projects) return 'loading...'
return 'You have ' + projects.length + ' projects'
}
```
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Data Fetching
type: guide
summary: Fetch data with SWR using custom fetcher functions.
prerequisites:
- /docs/getting-started
related:
- /docs/error-handling
- /docs/arguments
---
# Data Fetching
```js
const { data, error } = useSWR(key, fetcher)
```
This is the very fundamental API of SWR. The `fetcher` here is an async function that **accepts the `key`** of SWR, and returns the data.
The returned value will be passed as `data`, and if it throws, it will be caught as `error`.
Note that fetcher can be omitted from the parameters if it's{' '}
provided globally.
## Fetch
You can use any library to handle data fetching, for example a `fetch` polyfill [developit/unfetch](https://github.com/developit/unfetch):
```js
import fetch from 'unfetch'
const fetcher = url => fetch(url).then(r => r.json())
function App () {
const { data, error } = useSWR('/api/data', fetcher)
// ...
}
```
If you are using Next.js, you don't need to import this polyfill: New Built-In Polyfills: fetch(), URL, and Object.assign
## Axios
```js
import axios from 'axios'
const fetcher = url => axios.get(url).then(res => res.data)
function App () {
const { data, error } = useSWR('/api/data', fetcher)
// ...
}
```
## GraphQL
Or using GraphQL with libs like [graphql-request](https://github.com/prisma-labs/graphql-request):
```js
import { request } from 'graphql-request'
const fetcher = query => request('/api/graphql', query)
function App () {
const { data, error } = useSWR(
`{
Movie(title: "Inception") {
releaseDate
actors {
name
}
}
}`,
fetcher
)
// ...
}
```
*If you want to pass variables to a GraphQL query, check out [Multiple Arguments](/docs/arguments).*
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Error Handling
type: guide
summary: Handle errors when fetching data with SWR.
prerequisites:
- /docs/getting-started
related:
- /docs/data-fetching
- /docs/revalidation
---
# Error Handling
If an error is thrown inside [`fetcher`](/docs/data-fetching), it will be returned as `error` by the hook.
```js
const fetcher = url => fetch(url).then(r => r.json())
// ...
const { data, error } = useSWR('/api/user', fetcher)
```
The `error` object will be defined if the fetch promise is rejected.
## Status Code and Error Object
Sometimes we want an API to return an error object alongside the status code.
Both of them are useful for the client.
We can customize our `fetcher` to return more information. If the status code is not `2xx`,
we consider it an error even if it can be parsed as JSON:
```js
const fetcher = async url => {
const res = await fetch(url)
// If the status code is not in the range 200-299,
// we still try to parse and throw it.
if (!res.ok) {
const error = new Error('An error occurred while fetching the data.')
// Attach extra info to the error object.
error.info = await res.json()
error.status = res.status
throw error
}
return res.json()
}
// ...
const { data, error } = useSWR('/api/user', fetcher)
// error.info === {
// message: "You are not authorized to access this resource.",
// documentation_url: "..."
// }
// error.status === 403
```
Note that data and error can exist at the same time. So the UI can display the existing data,
while knowing the upcoming request has failed.
[Here](/examples/error-handling) we have an example.
## Error Retry
SWR uses the [exponential backoff algorithm](https://en.wikipedia.org/wiki/Exponential_backoff) to retry the request on error.
The algorithm allows the app to recover from errors quickly, but not waste resources retrying too often.
You can also override this behavior via the [onErrorRetry](/docs/api#options) option:
```js
useSWR('/api/user', fetcher, {
onErrorRetry: (error, key, config, revalidate, { retryCount }) => {
// Never retry on 404.
if (error.status === 404) return
// Never retry for a specific key.
if (key === '/api/user') return
// Only retry up to 10 times.
if (retryCount >= 10) return
// Retry after 5 seconds.
setTimeout(() => revalidate({ retryCount }), 5000)
}
})
```
This callback gives you the flexibility to retry based on various conditions. You can also disable it by setting `shouldRetryOnError: false`.
It's also possible to provide it via the [Global Configuration](/docs/global-configuration) context.
## Global Error Report
You can always get the `error` object inside the component reactively.
But in case you want to handle the error globally, to notify the UI to show a [toast](https://vercel.com/design/toast) or a [snackbar](https://material.io/components/snackbars), or report it somewhere such as [Sentry](https://sentry.io),
there's an [`onError`](/docs/api#options) event:
```jsx
{
if (error.status !== 403 && error.status !== 404) {
// We can send the error to Sentry,
// or show a notification UI.
}
}
}}>
```
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Getting Started
type: guide
summary: Install and set up SWR for data fetching in React applications.
related:
- /docs/data-fetching
- /docs/api
---
# Getting Started
## Installation
Inside your React project directory, run the following:
npm
pnpm
yarn
bun
```bash
npm i swr
```
```bash
pnpm add swr
```
```bash
yarn add swr
```
```bash
bun add swr
```
## Quick Start
For normal RESTful APIs with JSON data, first you need to create a `fetcher` function, which is just a wrapper of the native `fetch`:
```jsx
const fetcher = (...args) => fetch(...args).then(res => res.json())
```
If you want to use GraphQL API or libs like Axios, you can create your own fetcher function.
Check here for more examples.
Then you can import `useSWR` and start using it inside any function components:
```jsx
import useSWR from 'swr'
function Profile ({ userId }) {
const { data, error, isLoading } = useSWR(`/api/user/${userId}`, fetcher)
if (error) return
failed to load
if (isLoading) return
loading...
// render data
return
hello {data.name}!
}
```
Normally, there're 3 possible states of a request: "loading", "ready", or "error". You can use the value of `data`, `error` and `isLoading` to
determine the current state of the request, and return the corresponding UI.
## Make It Reusable
When building a web app, you might need to reuse the data in many places of the UI. It is incredibly easy to create reusable data hooks
on top of SWR:
```jsx
function useUser (id) {
const { data, error, isLoading } = useSWR(`/api/user/${id}`, fetcher)
return {
user: data,
isLoading,
isError: error
}
}
```
And use it in your components:
```jsx
function Avatar ({ userId }) {
const { user, isLoading, isError } = useUser(userId)
if (isLoading) return
if (isError) return
return
}
```
By adopting this pattern, you can forget about **fetching** data in the imperative way: start the request, update the loading state, and return the final result.
Instead, your code is more declarative: you just need to specify what data is used by the component.
## Example
In a real-world example, our website shows a navbar and the content, both depend on `user`:
Traditionally, we fetch data once using `useEffect` in the top level component, and pass it to child components via props (notice that we don't handle error state for now):
```jsx {7-11,17,18,27}
// page component
function Page ({ userId }) {
const [user, setUser] = useState(null)
// fetch data
useEffect(() => {
fetch(`/api/user/${userId}`)
.then(res => res.json())
.then(data => setUser(data))
}, [userId])
// global loading state
if (!user) return
return
}
// child components
function Navbar ({ user }) {
return
...
}
function Content ({ user }) {
return
Welcome back, {user.name}
}
function Avatar ({ user }) {
return
}
```
Usually, we need to keep all the data fetching in the top level component and add props to every component deep down the tree.
The code will become harder to maintain if we add more data dependency to the page.
Although we can avoid passing props using [Context](https://react.dev/learn/passing-data-deeply-with-context), there's still the dynamic content problem:
components inside the page content can be dynamic, and the top level component might not know what data will be needed by its child components.
SWR solves the problem perfectly. With the `useUser` hook we just created, the code can be refactored to:
```jsx {20,26}
// page component
function Page ({ userId }) {
return
}
function Content ({ userId }) {
const { user, isLoading } = useUser(userId)
if (isLoading) return
return
Welcome back, {user.name}
}
function Avatar ({ userId }) {
const { user, isLoading } = useUser(userId)
if (isLoading) return
return
}
```
Data is now **bound** to the components which need the data, and all components are **independent** to each other.
All the parent components don't need to know anything about the data or passing data around. They just render.
The code is much simpler and easier to maintain now.
The most beautiful thing is that there will be only **1 request** sent to the API, because they use the same SWR key and
the request is **deduped**, **cached** and **shared** automatically.
Also, the application now has the ability to refetch the data on [user focus or network reconnect](/docs/revalidation)!
That means, when the user's laptop wakes from sleep or they switch between browser tabs, the data will be refreshed automatically.
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Global Configuration
type: reference
summary: Configure default options for all SWR hooks.
prerequisites:
- /docs/getting-started
related:
- /docs/api
- /docs/middleware
---
# Global Configuration
The context `SWRConfig` can provide global configurations ([options](/docs/api)) for all SWR hooks.
```jsx
```
In this example, all SWR hooks will use the same fetcher provided to load JSON data, and refresh every 3 seconds by default:
```jsx
import useSWR, { SWRConfig } from 'swr'
function Dashboard () {
const { data: events } = useSWR('/api/events')
const { data: projects } = useSWR('/api/projects')
const { data: user } = useSWR('/api/user', { refreshInterval: 0 }) // override
// ...
}
function App () {
return (
fetch(resource, init).then(res => res.json())
}}
>
)
}
```
## Nesting Configurations
`SWRConfig` merges the configuration from the parent context. It can receive either an object or a functional configuration. The functional one receives the parent configuration as argument and returns a new configuration that you can customize yourself.
### Object Configuration Example
```jsx
import { SWRConfig, useSWRConfig } from 'swr'
function App() {
return (
)
}
function Page() {
const config = useSWRConfig()
// {
// dedupingInterval: 200,
// refreshInterval: 100,
// fallback: { a: 2, b: 1, c: 2 },
// }
}
```
### Functional Configuration Example
```jsx
import { SWRConfig, useSWRConfig } from 'swr'
function App() {
return (
({
dedupingInterval: parent.dedupingInterval * 5,
fallback: { a: 2, c: 2 },
})}
>
)
}
function Page() {
const config = useSWRConfig()
// {
// dedupingInterval: 500,
// fallback: { a: 2, c: 2 },
// }
}
```
## Extra APIs
### Cache Provider
Besides all the [options](/docs/api) listed, `SWRConfig` also accepts an optional `provider` function. Please refer to the [Cache](/docs/advanced/cache) section for more details.
```jsx
new Map() }}>
```
### Access To Global Configurations
You can use the `useSWRConfig` hook to get the global configurations, as well as [`mutate`](/docs/mutation), [`cache`](/docs/advanced/cache), and the provider-scoped [`unload`](/docs/advanced/cache#clear-the-cache) function:
```jsx
import { useSWRConfig } from 'swr'
function Component () {
const { refreshInterval, mutate, cache, unload, ...restConfig } = useSWRConfig()
// ...
}
```
Nested configurations will be extended. If no `` is used, it will return the default ones.
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Middleware
type: conceptual
summary: Extend SWR with custom middleware.
prerequisites:
- /docs/getting-started
related:
- /docs/global-configuration
---
# Middleware
Upgrade to the latest version (≥ 1.0.0) to use this feature.
The middleware feature is a new addition in SWR 1.0 that enables you to execute logic before and after SWR hooks.
## Usage
Middleware receive the SWR hook and can execute logic before and after running it. If there are multiple middleware, each middleware wraps the next middleware. The last middleware in the list will receive the original SWR hook `useSWR`.
### API
*Notes: The function name shouldn't be capitalized (e.g. `myMiddleware` instead of `MyMiddleware`) or React lint rules will throw `Rules of Hook` error*
[TypeScript](https://swr.vercel.app/docs/typescript#middleware-types)
```jsx
function myMiddleware (useSWRNext) {
return (key, fetcher, config) => {
// Before hook runs...
// Handle the next middleware, or the `useSWR` hook if this is the last one.
const swr = useSWRNext(key, fetcher, config)
// After hook runs...
return swr
}
}
```
You can pass an array of middleware as an option to `SWRConfig` or `useSWR`:
```jsx
// or...
useSWR(key, fetcher, { use: [myMiddleware] })
```
### Extend
Middleware will be extended like regular options. For example:
```jsx
function Bar () {
useSWR(key, fetcher, { use: [c] })
// ...
}
function Foo() {
return (
)
}
```
is equivalent to:
```js
useSWR(key, fetcher, { use: [a, b, c] })
```
### Multiple Middleware
Each middleware wraps the next middleware, and the last one just wraps the SWR hook. For example:
```jsx
useSWR(key, fetcher, { use: [a, b, c] })
```
The order of middleware executions will be `a → b → c`, as shown below:
```plaintext
enter a
enter b
enter c
useSWR()
exit c
exit b
exit a
```
## Examples
### Request Logger
Let's build a simple request logger middleware as an example. It prints out all the fetcher requests sent from this SWR hook. You can also use this middleware for all SWR hooks by adding it to `SWRConfig`.
```jsx
function logger(useSWRNext) {
return (key, fetcher, config) => {
// Add logger to the original fetcher.
const extendedFetcher = (...args) => {
console.log('SWR Request:', key)
return fetcher(...args)
}
// Execute the hook with the new fetcher.
return useSWRNext(key, extendedFetcher, config)
}
}
// ... inside your component
useSWR(key, fetcher, { use: [logger] })
```
Every time the request is fired, it outputs the SWR key to the console:
```plaintext
SWR Request: /api/user1
SWR Request: /api/user2
```
### Keep Previous Result
Sometimes you want the data returned by `useSWR` to be "laggy". Even if the key changes,
you still want it to return the previous result until the new data has loaded.
This can be built as a laggy middleware together with `useRef`. In this example, we are also going to
extend the returned object of the `useSWR` hook:
```jsx
import { useRef, useEffect, useCallback } from 'react'
// This is a SWR middleware for keeping the data even if key changes.
function laggy(useSWRNext) {
return (key, fetcher, config) => {
// Use a ref to store previous returned data.
const laggyDataRef = useRef()
// Actual SWR hook.
const swr = useSWRNext(key, fetcher, config)
useEffect(() => {
// Update ref if data is not undefined.
if (swr.data !== undefined) {
laggyDataRef.current = swr.data
}
}, [swr.data])
// Expose a method to clear the laggy data, if any.
const resetLaggy = useCallback(() => {
laggyDataRef.current = undefined
}, [])
// Fallback to previous data if the current data is undefined.
const dataOrLaggyData = swr.data === undefined ? laggyDataRef.current : swr.data
// Is it showing previous data?
const isLagging = swr.data === undefined && laggyDataRef.current !== undefined
// Also add a `isLagging` field to SWR.
return Object.assign({}, swr, {
data: dataOrLaggyData,
isLagging,
resetLaggy,
})
}
}
```
When you need a SWR hook to be laggy, you can then use this middleware:
```js
const { data, isLagging, resetLaggy } = useSWR(key, fetcher, { use: [laggy] })
```
### Serialize Object Keys
Since SWR 1.1.0, object-like keys will be serialized under the hood automatically.
In older versions (\< 1.1.0), SWR **shallowly** compares the arguments on every render, and triggers revalidation if any of them has changed.
If you are passing serializable objects as the key. You can serialize object keys to ensure its stability, a simple middleware can help:
```jsx
function serialize(useSWRNext) {
return (key, fetcher, config) => {
// Serialize the key.
const serializedKey = Array.isArray(key) ? JSON.stringify(key) : key
// Pass the serialized key, and unserialize it in fetcher.
return useSWRNext(serializedKey, (k) => fetcher(...JSON.parse(k)), config)
}
}
// ...
useSWR(['/api/user', { id: '73' }], fetcher, { use: [serialize] })
// ... or enable it globally with
```
You don’t need to worry that object might change between renders. It’s always serialized to the same string, and the fetcher will still receive those object arguments.
Furthermore, you can use libs like [fast-json-stable-stringify](https://github.com/epoberezkin/fast-json-stable-stringify) instead of `JSON.stringify` — faster and stabler.
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Mutation & Revalidation
type: guide
summary: Mutate cached data and trigger revalidation.
prerequisites:
- /docs/data-fetching
related:
- /docs/data-fetching
- /docs/revalidation
---
# Mutation & Revalidation
SWR provides the [`mutate`](/docs/mutation#mutate) and [`useSWRMutation`](/docs/mutation#useswrmutation) APIs for mutating remote data and related cache.
## `mutate`
There're 2 ways to use the `mutate` API to mutate the data, the global mutate API which can mutate any key and the bound mutate API which only can mutate the data of corresponding SWR hook.
#### Global Mutate
The recommended way to get the global mutator is to use the [`useSWRConfig`](/docs/global-configuration#access-to-global-configurations) hook:
```js
import { useSWRConfig } from "swr"
function App() {
const { mutate } = useSWRConfig()
mutate(key, data, options)
}
```
You can also import it globally:
```js
import { mutate } from "swr"
function App() {
mutate(key, data, options)
}
```
Using global mutator only with the `key` parameter will ***not update the cache or trigger revalidation*** unless there is a mounted SWR hook using the same key.
#### Bound Mutate
Bound mutate is the short path to mutate the current key with data. Which `key` is bounded to the `key` passing to `useSWR`, and receive the `data` as the first argument.
It is functionally equivalent to the global `mutate` function in the previous section but does not require the `key` parameter:
```jsx
import useSWR from 'swr'
function Profile () {
const { data, mutate } = useSWR('/api/user', fetcher)
return (
My name is {data.name}.
)
}
```
#### Revalidation
When you call `mutate(key)` (or just `mutate()` with the bound mutate API) without any data, it will trigger a revalidation (mark the data as expired and trigger a refetch)
for the resource. This example shows how to automatically refetch the login info (e.g. inside ``)
when the user clicks the “Logout” button:
```jsx {14}
import useSWR, { useSWRConfig } from 'swr'
function App () {
const { mutate } = useSWRConfig()
return (
)
}
```
It broadcasts to SWR hooks under the same [cache provider](/docs/advanced/cache) scope. If no cache provider exists, it will broadcast to all SWR hooks.
### API
#### Parameters
* `key`: same as `useSWR`'s `key`, but a function behaves as [a filter function](/docs/mutation#mutate-multiple-items)
* `data`: data to update the client cache, or an async function for the remote mutation
* `options`: accepts the following options
* `optimisticData`: data to immediately update the client cache, or a function that receives current data and returns the new client cache data, usually used in optimistic UI.
* `revalidate = true`: should the cache revalidate once the asynchronous update resolves. If set to a function, the function receives `data` and `key`.
* `populateCache = true`: should the result of the remote mutation be written to the cache, or a function that receives new result and current result as arguments and returns the mutation result.
* `rollbackOnError = true`: should the cache rollback if the remote mutation errors, or a function that receives the error thrown from fetcher as arguments and returns a boolean whether should rollback or not.
* `throwOnError = true`: should the mutate call throw the error when fails.
#### Return Values
`mutate` returns the results the `data` parameter has been resolved. The function passed to `mutate` will return an updated data which is used to update the corresponding cache value. If there is an error thrown while executing the function, the error will be thrown so it can be handled appropriately.
```jsx
try {
const user = await mutate('/api/user', updateUser(newUser))
} catch (error) {
// Handle an error while updating the user here
}
```
## `useSWRMutation`
SWR also provides `useSWRMutation` as a hook for remote mutations. The remote mutations are only triggered manually, instead of automatically like `useSWR`.
Also, this hook doesn’t share states with other `useSWRMutation` hooks.
```tsx
import useSWRMutation from 'swr/mutation'
// Fetcher implementation.
// The extra argument will be passed via the `arg` property of the 2nd parameter.
// In the example below, `arg` will be `'my_token'`
async function updateUser(url, { arg }: { arg: string }) {
await fetch(url, {
method: 'POST',
headers: {
Authorization: `Bearer ${arg}`
}
})
}
function Profile() {
// A useSWR + mutate like API, but it will not start the request automatically.
const { trigger } = useSWRMutation('/api/user', updateUser, options)
return
}
```
### API
#### Parameters
* `key`: same as [`mutate`](/docs/mutation#mutate)'s `key`
* `fetcher(key, { arg })`: an async function for remote mutation
* `options`: an optional object with the following properties:
* `optimisticData`: same as `mutate`'s `optimisticData`
* `revalidate = true`: same as `mutate`'s `revalidate`
* `populateCache = false`: same as `mutate`'s `populateCache`, but the default is `false`
* `rollbackOnError = true`: same as `mutate`'s `rollbackOnError`
* `throwOnError = true`: same as `mutate`'s `throwOnError`
* `onSuccess(data, key, config)`:　 callback function when a remote mutation has been finished successfully
* `onError(err, key, config)`: callback function when a remote mutation has returned an error
#### Return Values
* `data`: data for the given key returned from `fetcher`
* `error`: error thrown by `fetcher` (or undefined)
* `trigger(arg, options)`: a function to trigger a remote mutation
* `reset`: a function to reset the state (`data`, `error`, `isMutating`)
* `isMutating`: if there's an ongoing remote mutation
### Basic Usage
```tsx
import useSWRMutation from 'swr/mutation'
async function sendRequest(url, { arg }: { arg: { username: string }}) {
return fetch(url, {
method: 'POST',
body: JSON.stringify(arg)
}).then(res => res.json())
}
function App() {
const { trigger, isMutating } = useSWRMutation('/api/user', sendRequest, /* options */)
return (
)
}
```
If you want to use the mutation results in rendering, you can get them from the return values of `useSWRMutation`.
```jsx
const { trigger, data, error } = useSWRMutation('/api/user', sendRequest)
```
`useSWRMutation` shares a cache store with `useSWR`, so it can detect and avoid race conditions between `useSWR`. It also supports `mutate`'s functionalities like optimistic updates and rollback on errors. You can pass these options `useSWRMutation` and its `trigger` function.
```jsx
const { trigger } = useSWRMutation('/api/user', updateUser, {
optimisticData: current => ({ ...current, name: newName })
})
// or
trigger(newName, {
optimisticData: current => ({ ...current, name: newName })
})
```
### Defer loading data until needed
You can also use `useSWRMutation` for loading data. `useSWRMutation` won't start requesting until `trigger` is called, so you can defer loading data when you actually need it.
```jsx
import { useState } from 'react'
import useSWRMutation from 'swr/mutation'
const fetcher = url => fetch(url).then(res => res.json())
const Page = () => {
const [show, setShow] = useState(false)
// data is undefined until trigger is called
const { data: user, trigger } = useSWRMutation('/api/user', fetcher);
return (
{show && user ?
{user.name}
: null}
);
}
```
## Optimistic Updates
In many cases, applying local mutations to data is a good way to make changes
feel faster — no need to wait for the remote source of data.
With the `optimisticData` option, you can update your local data manually, while
waiting for the remote mutation to finish. Composing `rollbackOnError` you can also
control when to rollback the data.
```jsx
import useSWR, { useSWRConfig } from 'swr'
function Profile () {
const { mutate } = useSWRConfig()
const { data } = useSWR('/api/user', fetcher)
return (
My name is {data.name}.
)
}
```
> The **`updateFn`** should be a promise or asynchronous function to handle the remote mutation, it should return updated data.
You can also pass a function to `optimisticData` to make it depending on the current data:
```jsx
import useSWR, { useSWRConfig } from 'swr'
function Profile () {
const { mutate } = useSWRConfig()
const { data } = useSWR('/api/user', fetcher)
return (
My name is {data.name}.
)
}
```
You can also create the same thing with `useSWRMutation` and `trigger`:
```jsx
import useSWRMutation from 'swr/mutation'
function Profile () {
const { trigger } = useSWRMutation('/api/user', updateUserName)
return (
My name is {data.name}.
)
}
```
## Rollback on Errors
When you have `optimisticData` set, it’s possible that the optimistic data gets
displayed to the user, but the remote mutation fails. In this case, you can leverage
`rollbackOnError` to revert the local cache to the previous state, to make sure
the user is seeing the correct data.
## Update Cache After Mutation
Sometimes, the remote mutation request directly returns the updated data, so there is no need to do an extra fetch to load it.
You can enable the `populateCache` option to update the cache for `useSWR` with the response of the mutation:
```jsx
const updateTodo = () => fetch('/api/todos/1', {
method: 'PATCH',
body: JSON.stringify({ completed: true })
})
mutate('/api/todos', updateTodo, {
populateCache: (updatedTodo, todos) => {
// filter the list, and return it with the updated item
const filteredTodos = todos.filter(todo => todo.id !== '1')
return [...filteredTodos, updatedTodo]
},
// Since the API already gives us the updated information,
// we don't need to revalidate here.
revalidate: false
})
```
Or with the `useSWRMutation` hook:
```jsx
useSWRMutation('/api/todos', updateTodo, {
populateCache: (updatedTodo, todos) => {
// filter the list, and return it with the updated item
const filteredTodos = todos.filter(todo => todo.id !== '1')
return [...filteredTodos, updatedTodo]
},
// Since the API already gives us the updated information,
// we don't need to revalidate here.
revalidate: false
})
```
When combined with `optimisticData` and `rollbackOnError`, you’ll get a perfect optimistic UI experience.
## Avoid Race Conditions
Both `mutate` and `useSWRMutation` can avoid race conditions between `useSWR`. For example,
```tsx
function Profile() {
const { data } = useSWR('/api/user', getUser, { revalidateInterval: 3000 })
const { trigger } = useSWRMutation('/api/user', updateUser)
return <>
{data ? data.username : null}
}
```
The normal `useSWR` hook might refresh its data any time due to focus, polling, or other conditions. This way the displayed username
can be as fresh as possible. However, since we have a mutation there that can happen at the nearly same time of a refetch of `useSWR`, there
could be a race condition that `getUser` request starts earlier, but takes longer than `updateUser`.
Luckily, `useSWRMutation` handles this for you automatically. After the mutation, it will tell `useSWR` to ditch the ongoing request and revalidate,
so the stale data will never be displayed.
## Mutate Based on Current Data
Sometimes, you want to update a part of your data based on the current data.
With `mutate`, you can pass an async function which will receive the current cached value, if any, and returns an updated document.
```jsx
mutate('/api/todos', async todos => {
// let's update the todo with ID `1` to be completed,
// this API returns the updated data
const updatedTodo = await fetch('/api/todos/1', {
method: 'PATCH',
body: JSON.stringify({ completed: true })
})
// filter the list, and return it with the updated item
const filteredTodos = todos.filter(todo => todo.id !== '1')
return [...filteredTodos, updatedTodo]
// Since the API already gives us the updated information,
// we don't need to revalidate here.
}, { revalidate: false })
```
## Mutate Multiple Items
The global `mutate` API accepts a filter function, which accepts `key` as the argument and returns which keys to revalidate. The filter function is applied to all the existing cache keys:
```jsx
import { mutate } from 'swr'
// Or from the hook if you customized the cache provider:
// { mutate } = useSWRConfig()
mutate(
key => typeof key === 'string' && key.startsWith('/api/item?id='),
undefined,
{ revalidate: true }
)
```
This also works with any key type like an array. The mutation matches all keys, of which the first element is `'item'`.
```jsx
useSWR(['item', 123], ...)
useSWR(['item', 124], ...)
useSWR(['item', 125], ...)
mutate(
key => Array.isArray(key) && key[0] === 'item',
undefined,
{ revalidate: false }
)
```
The filter function is applied to all existing cache keys, so you should not assume the shape of keys when using multiple shapes of keys.
```jsx
// âś… matching array key
mutate((key) => key[0].startsWith('/api'), data)
// âś… matching string key
mutate((key) => typeof key === 'string' && key.startsWith('/api'), data)
// ❌ ERROR: mutate uncertain keys (array or string)
mutate((key: any) => /\/api/.test(key.toString()))
```
You can use the filter function to clear all cache data, which is useful when logging out:
```js
const clearCache = () => mutate(
() => true,
undefined,
{ revalidate: false }
)
// ...clear cache on logout
clearCache()
```
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Pagination
type: guide
summary: Implement pagination and infinite loading patterns with SWR.
prerequisites:
- /docs/data-fetching
related:
- /docs/data-fetching
- /docs/mutation
---
# Pagination
Please update to the latest version (≥ 0.3.0) to use this API. The previous useSWRPages API is now deprecated.
SWR provides a dedicated API `useSWRInfinite` to support common UI patterns such as **pagination** and **infinite loading**.
## When to Use `useSWR`
### Pagination
First of all, we might **NOT** need `useSWRInfinite` but can use just `useSWR` if we are building something like this:
...which is a typical pagination UI. Let's see how it can be easily implemented with
`useSWR`:
```jsx {5}
function App () {
const [pageIndex, setPageIndex] = useState(0);
// The API URL includes the page index, which is a React state.
const { data } = useSWR(`/api/data?page=${pageIndex}`, fetcher);
// ... handle loading and error states
return
{data.map(item =>
{item.name}
)}
}
```
Furthermore, we can create an abstraction for this "page component":
```jsx {13}
function Page ({ index }) {
const { data } = useSWR(`/api/data?page=${index}`, fetcher);
// ... handle loading and error states
return data.map(item =>
}
```
Because of SWR's cache, we get the benefit to preload the next page. We render the next page inside
a hidden div, so SWR will trigger the data fetching of the next page. When the user navigates to the next page, the data is already there:
```jsx {6}
function App () {
const [pageIndex, setPageIndex] = useState(0);
return
}
```
With just 1 line of code, we get a much better UX. The `useSWR` hook is so powerful,
that most scenarios are covered by it.
### Infinite Loading
Sometimes we want to build an **infinite loading** UI, with a "Load More" button that appends data
to the list (or done automatically when you scroll):
To implement this, we need to make **dynamic number of requests** on this page. React Hooks have [a couple of rules](https://react.dev/reference/rules/rules-of-hooks),
so we **CANNOT** do something like this:
```jsx {5,6,7,8,9}
function App () {
const [cnt, setCnt] = useState(1)
const list = []
for (let i = 0; i < cnt; i++) {
// 🚨 This is wrong! Commonly, you can't use hooks inside a loop.
const { data } = useSWR(`/api/data?page=${i}`)
list.push(data)
}
return
{list.map((data, i) =>
{
data.map(item =>
{item.name}
)
}
)}
}
```
Instead, we can use the `` abstraction that we created to achieve it:
```jsx {5,6,7}
function App () {
const [cnt, setCnt] = useState(1)
const pages = []
for (let i = 0; i < cnt; i++) {
pages.push()
}
return
{pages}
}
```
### Advanced Cases
However, in some advanced use cases, the solution above doesn't work.
For example, we are still implementing the same "Load More" UI, but also need to display a number
about how many items are there in total. We can't use the `` solution anymore because
the top level UI (``) needs the data inside each page:
```jsx {10}
function App () {
const [cnt, setCnt] = useState(1)
const pages = []
for (let i = 0; i < cnt; i++) {
pages.push()
}
return
??? items
{pages}
}
```
Also, if the pagination API is **cursor based**, that solution doesn't work either. Because each page
needs the data from the previous page, they're not isolated.
That's how this new `useSWRInfinite` Hook can help.
## useSWRInfinite
`useSWRInfinite` gives us the ability to trigger a number of requests with one Hook. This is how it looks:
```jsx
import useSWRInfinite from 'swr/infinite'
// ...
const { data, error, isLoading, isValidating, mutate, size, setSize } = useSWRInfinite(
getKey, fetcher?, options?
)
```
Similar to `useSWR`, this new Hook accepts a function that returns the request key, a fetcher function, and options.
It returns all the values that `useSWR` returns, including 2 extra values: the page size and a page size setter, like a React state.
In infinite loading, one *page* is one request, and our goal is to fetch multiple pages and render them.
If you are using SWR 0.x versions, `useSWRInfinite` needs to be imported from `swr`:
`import { useSWRInfinite } from 'swr'`
### API
#### Parameters
* `getKey`: a function that accepts the index and the previous page data, returns the key of a page
* `fetcher`: same as `useSWR`'s [fetcher function](/docs/data-fetching)
* `options`: accepts all the options that `useSWR` supports, with 4 extra options:
* `initialSize = 1`: number of pages should be loaded initially
* `revalidateAll = false`: always try to revalidate all pages
* `revalidateFirstPage = true`: always try to revalidate the first page
* `persistSize = false`: don't reset the page size to 1 (or `initialSize` if set) when the first page's key changes
* `parallel = false`: fetches multiple pages in parallel
Note that the `initialSize` option is not allowed to change in the lifecycle.
#### Return Values
* `data`: an array of fetch response values of each page
* `error`: same as `useSWR`'s `error`
* `isLoading`: same as `useSWR`'s `isLoading`
* `isValidating`: same as `useSWR`'s `isValidating`
* `mutate`: same as `useSWR`'s bound mutate function but manipulates the data array
* `size`: the number of pages that *will* be fetched and returned
* `setSize`: set the number of pages that need to be fetched
### Example 1: Index Based Paginated API
For normal index based APIs:
```plaintext
GET /users?page=0&limit=10
[
{ name: 'Alice', ... },
{ name: 'Bob', ... },
{ name: 'Cathy', ... },
...
]
```
```jsx {4,5,6,7,10}
// A function to get the SWR key of each page,
// its return value will be accepted by `fetcher`.
// If `null` is returned, the request of that page won't start.
const getKey = (pageIndex, previousPageData) => {
if (previousPageData && !previousPageData.length) return null // reached the end
return `/users?page=${pageIndex}&limit=10` // SWR key
}
function App () {
const { data, size, setSize } = useSWRInfinite(getKey, fetcher)
if (!data) return 'loading'
// We can now calculate the number of all users
let totalUsers = 0
for (let i = 0; i < data.length; i++) {
totalUsers += data[i].length
}
return
{totalUsers} users listed
{data.map((users, index) => {
// `data` is an array of each page's API response.
return users.map(user =>
{user.name}
)
})}
}
```
The `getKey` function is the major difference between `useSWRInfinite` and `useSWR`.
It accepts the index of the current page, as well as the data from the previous page.
So both index based and cursor based pagination API can be supported nicely.
Also `data` is no longer just one API response. It's an array of multiple API responses:
```js
// `data` will look like this
[
[
{ name: 'Alice', ... },
{ name: 'Bob', ... },
{ name: 'Cathy', ... },
...
],
[
{ name: 'John', ... },
{ name: 'Paul', ... },
{ name: 'George', ... },
...
],
...
]
```
### Example 2: Cursor or Offset Based Paginated API
Let's say the API now requires a cursor and returns the next cursor alongside with the data:
```plaintext
GET /users?cursor=123&limit=10
{
data: [
{ name: 'Alice' },
{ name: 'Bob' },
{ name: 'Cathy' },
...
],
nextCursor: 456
}
```
We can change our `getKey` function to:
```jsx
const getKey = (pageIndex, previousPageData) => {
// reached the end
if (previousPageData && !previousPageData.data) return null
// first page, we don't have `previousPageData`
if (pageIndex === 0) return `/users?limit=10`
// add the cursor to the API endpoint
return `/users?cursor=${previousPageData.nextCursor}&limit=10`
}
```
### Parallel Fetching Mode
Please update to the latest version (≥ 2.1.0) to use this API.
The default behavior of useSWRInfinite is to fetch data for each page in sequence, as key creation is based on the previously fetched data. However, fetching data sequentially for a large number of pages may not be optimal, particularly if the pages are not interdependent. By specifying `parallel` option to `true` will let you fetch pages independently in parallel, which can significantly speed up the loading process.
```jsx
// parallel = false (default)
// page1 ===> page2 ===> page3 ===> done
//
// parallel = true
// page1 ==> done
// page2 =====> done
// page3 ===> done
//
// previousPageData is always `null`
const getKey = (pageIndex, previousPageData) => {
return `/users?page=${pageIndex}&limit=10`
}
function App () {
const { data } = useSWRInfinite(getKey, fetcher, { parallel: true })
}
```
The `previousPageData` argument of the `getKey` function becomes `null` when you enable the `parallel` option.
### Revalidate Specific Pages
Please update to the latest version (≥ 2.2.5) to use this API.
The default behavior of the mutation of `useSWRInfinite` is to revalidate all pages that have been loaded. But you might want to revalidate only the specific pages that have been changed. You can revalidate only specific pages by passing a function to the `revalidate` option.
The `revalidate` function is called for each page.
```jsx
function App() {
const { data, mutate, size } = useSWRInfinite(
(index) => [`/api/?page=${index + 1}`, index + 1],
fetcher
);
mutate(data, {
// only revalidate the last page
revalidate: (pageData, [url, page]) => page === size
});
}
```
### Global Mutate with `useSWRInfinite`
`useSWRInfinite` stores all page data into the cache with a special cache key along with each page data, so you have to use `unstable_serialize` in `swr/infinite` to revalidate the data with the global mutate.
```jsx
import { useSWRConfig } from "swr"
import { unstable_serialize } from "swr/infinite"
function App() {
const { mutate } = useSWRConfig()
mutate(unstable_serialize(getKey))
}
```
As the name implies, `unstable_serialize` is not a stable API, so we might change it in the future.
### Advanced Features
[Here is an example](/examples/infinite-loading) showing how you can implement the following features with `useSWRInfinite`:
* loading states
* show a special UI if it's empty
* disable the "Load More" button if reached the end
* changeable data source
* refresh the entire list
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Prefetching Data
type: guide
summary: Prefetch data to improve perceived performance.
prerequisites:
- /docs/data-fetching
related:
- /docs/data-fetching
- /docs/advanced/performance
---
# Prefetching Data
## Top-Level Page Data
There’re many ways to prefetch the data for SWR. For top level requests, [`rel="preload"`](https://developer.mozilla.org/en-US/docs/Web/HTML/Preloading_content) is highly recommended:
```html
```
Just put it inside your HTML ``. It’s easy, fast and native.
It will prefetch the data when the HTML loads, even before JavaScript starts to download. All your incoming fetch requests with the same URL will reuse the result (including SWR, of course).
## Programmatically Prefetch
SWR provides the `preload` API to prefetch the resources programmatically and store the results in the cache. `preload` accepts `key` and `fetcher` as the arguments.
You can call `preload` even outside of React.
```jsx
import { useState } from 'react'
import useSWR, { preload } from 'swr'
const fetcher = (url) => fetch(url).then((res) => res.json())
// Preload the resource before rendering the User component below,
// this prevents potential waterfalls in your application.
// You can also start preloading when hovering the button or link, too.
preload('/api/user', fetcher)
function User() {
const { data } = useSWR('/api/user', fetcher)
...
}
export default function App() {
const [show, setShow] = useState(false)
return (
{show ? : null}
)
}
```
Within React rendering tree, `preload` is also available to use in event handlers or effects.
```jsx
function App({ userId }) {
const [show, setShow] = useState(false)
// preload in effects
useEffect(() => {
preload('/api/user?id=' + userId, fetcher)
}, [userId])
return (
{show ? : null}
)
}
```
Together with techniques like [page prefetching](https://nextjs.org/docs/api-reference/next/router#routerprefetch) in Next.js, you will be able to load both next page and data instantly.
In Suspense mode, you should utilize `preload` to avoid waterfall problems.
```jsx
import useSWR, { preload } from 'swr'
// should call before rendering
preload('/api/user', fetcher);
preload('/api/movies', fetcher);
const Page = () => {
// The below useSWR hooks will suspend the rendering, but the requests to `/api/user` and `/api/movies` have started by `preload` already,
// so the waterfall problem doesn't happen.
const { data: user } = useSWR('/api/user', fetcher, { suspense: true });
const { data: movies } = useSWR('/api/movies', fetcher, { suspense: true });
return (
);
}
```
## Pre-fill Data
If you want to pre-fill existing data into the SWR cache, you can use the `fallbackData` option. For example:
```jsx
useSWR('/api/data', fetcher, { fallbackData: prefetchedData })
```
If SWR hasn't fetched the data yet, this hook will return `prefetchedData` as a fallback.
You can also configure this for all SWR hooks and multiple keys with `` and the `fallback` option. Check [Next.js SSG and SSR](/docs/with-nextjs) for more details.
## Pre-fetch in React Server Components (RSC)
When using Next.js App Router with React Server Components, you can prefetch data on the server and pass it to client components via `SWRConfig`. This approach allows you to leverage server-side data fetching while maintaining the benefits of SWR's client-side caching and revalidation.
For detailed information and examples, see the [Usage with Next.js](/docs/with-nextjs#prefetch-data-in-server-components) documentation.
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Automatic Revalidation
type: conceptual
summary: Automatically revalidate stale data with focus, interval, and reconnect strategies.
prerequisites:
- /docs/getting-started
related:
- /docs/mutation
- /docs/conditional-fetching
---
# Automatic Revalidation
If you want to manually revalidate the data, check mutation.
## Revalidate on Focus
When you re-focus a page or switch between tabs, SWR automatically revalidates data.
This can be useful to immediately synchronize to the latest state. This is helpful for refreshing data in scenarios like stale mobile tabs, or laptops that **went to sleep**.
This feature is enabled by default. You can disable it via the [`revalidateOnFocus`](/docs/api) option.
## Revalidate on Interval
In many cases, data changes because of multiple devices, multiple users, multiple tabs. How can we over time update the data on screen?
SWR will give you the option to automatically refetch data. It’s **smart** which means refetching will only happen if the component associated with the hook is **on screen**.
You can enable it by setting a [`refreshInterval`](/docs/api) value:
```js
useSWR('/api/todos', fetcher, { refreshInterval: 1000 })
```
There're also options such as `refreshWhenHidden` and `refreshWhenOffline`. Both are disabled by default so SWR won't fetch when the webpage is not on screen, or there's no network connection.
## Revalidate on Reconnect
It's useful to also revalidate when the user is back online. This scenario happens a lot when the user unlocks their computer, but the internet is not yet connected at the same moment.
To make sure the data is always up-to-date, SWR automatically revalidates when network recovers.
This feature is enabled by default. You can disable it via the [`revalidateOnReconnect`](/docs/api) option.
## Disable Automatic Revalidations
If the resource is **immutable**, that will never change if we revalidate again, we can disable all kinds of automatic revalidations for it.
Since version 1.0, SWR provides a helper hook `useSWRImmutable` to mark the resource as immutable:
```js
import useSWRImmutable from 'swr/immutable'
// ...
useSWRImmutable(key, fetcher, options)
```
It has the same API interface as the normal `useSWR` hook. You can also do the same thing by disabling the following revalidation options:
```js
useSWR(key, fetcher, {
revalidateIfStale: false,
revalidateOnFocus: false,
revalidateOnReconnect: false
})
// equivalent to
useSWRImmutable(key, fetcher)
```
The `revalidateIfStale` controls if SWR should revalidate when it mounts and there is stale data.
These 2 hooks above do the **exact same** thing. Once the data is cached, they will never request it again.
## Revalidate on Mount
It's useful to force override SWR revalidation on mounting. By default, the value of `revalidateOnMount` is set to undefined.
A SWR hook mounts as:
* First it checks if `revalidateOnMount` is defined. It starts request if it's true, stop if it's false.
`revalidateIfStale` useful to control the mount behaviour. By default `revalidateIfStale` is set to true.
If `revalidateIfStale` is set to true it only refetches if there's any cache data else it will not refetch.
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Subscription
type: guide
summary: Subscribe to real-time data sources with SWR.
prerequisites:
- /docs/getting-started
related:
- /docs/data-fetching
---
# Subscription
Please update to the latest version (≥ 2.1.0) to use this API.
## `useSWRSubscription`
`useSWRSubscription` is a React hook that allows subscribing to real-time data sources with SWR.
```tsx
useSWRSubscription(key: Key, subscribe: (key: Key, options: { next: (error?: Error | null, data: Data) => void }) => () => void): { data?: Data, error?: Error }
```
### API
This hook subscribes to a real-time data source using the subscribe function provided, and returns the latest data received and any errors encountered. The hook automatically updates the returned data as new events are received.
#### Parameters
* `key`: A unique key that identifies the data being subscribed to, same key as `useSWR` key.
* `subscribe`: A function that subscribes to the real-time data source. It receives the following arguments:
* `key`: same key as above
* `options`: an object with the following properties:
* `next`: A function that accepts an error and data, and updates the state with the latest data received from the real-time data source.
For instance
```tsx
function subscribe(key, { next }) {
const sub = remote.subscribe(key, (err, data) => next(err, data))
return () => sub.close()
}
```
You could also pass a updater function as `data` to `next`, which will receive the previous data as the first argument and return the new data.
```tsx
function subscribe(key, { next }) {
const sub = remote.subscribe(key, (err, data) => next(err, prev => prev.concat(data)))
return () => sub.close()
}
```
#### Return Values
* `state`: An object with the following properties:
* `data`: The latest data received from the real-time data source.
* `error`: An Error object if an error occurred while subscribing to the real-time data source, otherwise undefined.
When new data is received, the `error` will be reset to `undefined`.
### Usage
Using `useSWRSubscription` to subscribe to a Firestore data source:
```tsx
import useSWRSubscription from 'swr/subscription'
function Post({ id }) {
const { data } = useSWRSubscription(['views', id], ([_, postId], { next }) => {
const ref = firebase.database().ref('views/' + postId)
ref.on('value',
snapshot => next(null, snapshot.data()),
err => next(err)
)
return () => ref.off()
})
return Your post has {data} views!
}
```
Using `useSWRSubscription` to subscribe to a WebSocket data source:
```tsx
import useSWRSubscription from 'swr/subscription'
function App() {
const { data, error } = useSWRSubscription('ws://...', (key, { next }) => {
const socket = new WebSocket(key)
socket.addEventListener('message', (event) => next(null, event.data))
socket.addEventListener('error', (event) => next(event.error))
return () => socket.close()
})
if (error) return
failed to load
if (!data) return
loading...
return
hello {data}!
}
```
You could also check TypeScript examples of `useSWRSubscription` at [this page](/docs/typescript#useswrsubscription)
### Deduplication
`useSWRSubscription` deduplicates the subscription requests with the same key.
If there are multiple components using the same key, they will share the same subscription.
When the last component using the key unmounts, the subscription will be closed.
This means that if you have multiple components using the same key, they will all receive the same data.
And there's only one subscription to the real-time data source per key.
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Suspense
type: guide
summary: Use SWR with React Suspense for declarative data fetching.
prerequisites:
- /docs/getting-started
related:
- /docs/data-fetching
---
# Suspense
You can enable the `suspense` option to use SWR with React [Suspense](https://react.dev/reference/react/Suspense):
```jsx
import { Suspense } from 'react'
import useSWR from 'swr'
function Profile () {
const { data } = useSWR('/api/user', fetcher, { suspense: true })
return
hello, {data.name}
}
function App () {
return (
loading...}>
)
}
```
Note that the `suspense` option is not allowed to change in the lifecycle.
In Suspense mode, `data` is always the fetch response (so you don't need to check if it's `undefined`).
But if an error occurred, you need to use an [error boundary](https://reactjs.org/docs/concurrent-mode-suspense.html#handling-errors) to catch it:
```jsx
Could not fetch posts.}>
Loading posts...}>
```
Suspense mode suspends rendering until the data is ready, which means it causes waterfall problems easily. To avoid that, you should prefetch resources before rendering. [More information](/docs/prefetching)
***
### Note: With Conditional Fetching
Normally, when you enabled `suspense` it's guaranteed that `data` will always be ready on render:
```jsx
function Profile () {
const { data } = useSWR('/api/user', fetcher, { suspense: true })
// `data` will never be `undefined`
// ...
}
```
However, when using it together with conditional fetching or dependent fetching, `data` will be `undefined` if the request is **paused**:
```jsx
function Profile () {
const { data } = useSWR(isReady ? '/api/user' : null, fetcher, { suspense: true })
// `data` will be `undefined` if `isReady` is false
// ...
}
```
If you want to read more technical details about this restriction, check [the discussion here](https://github.com/vercel/swr/pull/357#issuecomment-627089889).
### Server-Side Rendering
When using suspense mode on the server-side (including pre-rendering in Next.js), it's **required** to provide the initial data via [fallbackData or fallback](/docs/with-nextjs#pre-rendering-with-default-data). This means that you can't use `Suspense` to fetch data on the server side, but either doing fully client-side data fetching, or fetch the data via the framework level data fetching method(such as getStaticProps in Next.js). More discussions can be found [here](https://github.com/vercel/swr/issues/1906).
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: TypeScript
type: reference
summary: Use SWR with TypeScript for type-safe data fetching.
prerequisites:
- /docs/getting-started
related:
- /docs/api
---
# TypeScript
SWR is friendly for apps written in TypeScript, with type safety out of the box.
## Basic Usage
By default, SWR will also infer the argument types of `fetcher` from `key`, so you can have the preferred types automatically.
### useSWR
```typescript
// `key` is inferred to be `string`
useSWR('/api/user', key => {})
useSWR(() => '/api/user', key => {})
// `key` will be inferred as { a: string; b: { c: string; d: number } }
useSWR({ a: '1', b: { c: '3', d: 2 } }, key => {})
useSWR(() => ({ a: '1', b: { c: '3', d: 2 } }), key => {})
// `arg0` will be inferred as string. `arg1` will be inferred as number
useSWR(['user', 8], ([arg0, arg1]) => {})
useSWR(() => ['user', 8], ([arg0, arg1]) => {})
```
You can also explicitly specify the types for `key` and `fetcher`'s arguments.
```typescript
import useSWR, { Fetcher } from 'swr'
const uid = ''
const fetcher: Fetcher = (id) => getUserById(id)
const { data } = useSWR(uid, fetcher)
// `data` will be `User | undefined`.
```
By default, [the error thrown](/docs/error-handling) inside the `fetcher` function has type `any`. The type can also be explicitly specified.
```typescript
const { data, error } = useSWR(uid, fetcher);
// `data` will be `User | undefined`.
// `error` will be `Error | undefined`.
```
### useSWRInfinite
Same for `swr/infinite`, you can either rely on the automatic type inference or explicitly specify the types by yourself.
```typescript
import { SWRInfiniteKeyLoader } from 'swr/infinite'
const getKey: SWRInfiniteKeyLoader = (index, previousPageData) => {
// ...
}
const { data } = useSWRInfinite(getKey, fetcher)
```
### useSWRSubscription
* Inline subscribe function and manually specify the type of `next` using `SWRSubscriptionOptions`.
```tsx
import useSWRSubscription from 'swr/subscription'
import type { SWRSubscriptionOptions } from 'swr/subscription'
const { data, error } = useSWRSubscription('key',
(key, { next }: SWRSubscriptionOptions) => {
//^ key will be inferred as `string`
//....
})
return {
data,
//^ data will be inferred as `number | undefined`
error
//^ error will be inferred as `Error | undefined`
}
}
```
* declare subscribe function using `SWRSubscription`
```tsx
import useSWRSubscription from 'swr/subscription'
import type { SWRSubscription } from 'swr/subscription'
/**
* The first generic is Key
* The second generic is Data
* The Third generic is Error
*/
const sub: SWRSubscription = (key, { next }) => {
//......
}
const { data, error } = useSWRSubscription('key', sub)
```
## Generics
Specifying the type of `data` is easy. By default, it will use the return type of `fetcher` (with `undefined` for the non-ready state) as the `data` type, but you can also pass it as a parameter:
```typescript
// 🔹 A. Use a typed fetcher:
// `getUser` is `(endpoint: string) => User`.
const { data } = useSWR('/api/user', getUser)
// 🔹 B. Specify the data type:
// `fetcher` is generally returning `any`.
const { data } = useSWR('/api/user', fetcher)
```
If you want to add types for other options of SWR, you can also import those types directly:
```typescript
import useSWR from 'swr'
import type { SWRConfiguration } from 'swr'
const config: SWRConfiguration = {
fallbackData: "fallback",
revalidateOnMount: false
// ...
}
const { data } = useSWR('/api/data', fetcher, config)
```
## Middleware Types
There're some extra type definitions you can import to help adding types to your custom middleware.
```typescript
import useSWR, { Middleware, SWRHook } from 'swr'
const swrMiddleware: Middleware = (useSWRNext: SWRHook) => (key, fetcher, config) => {
// ...
return useSWRNext(key, fetcher, config)
}
```
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Usage with Next.js
type: integration
summary: Integrate SWR with Next.js for server-side rendering and static generation.
prerequisites:
- /docs/getting-started
related:
- /docs/prefetching
- /docs/advanced/cache
---
# Usage with Next.js
## App Router
### Server Components
In Next.js App Router, all components are React Server Components (RSC) by default. **You can import SWRConfig and the key serialization APIs from SWR in RSC.**
```tsx filename="app/page.tsx" copy
import { unstable_serialize } from 'swr' // âś… Available in server components
import { unstable_serialize as infinite_unstable_serialize } from 'swr/infinite' // âś… Available in server components
import { SWRConfig } from 'swr' // âś… Available in server components
import { preload } from 'swr' // âś… Available in server components
```
You could not import hook APIs from SWR since they are not available in RSC.
```tsx filename="app/page.tsx" highlight={1}
import useSWR from 'swr' // ❌ This is not available in server components
import useSWRInfinite from 'swr/infinite' // ❌ This is not available in server components
import usesSWRMutation from 'swr/mutation' // ❌ This is not available in server components
```
### Client Components
You can mark your components with `'use client'` directive or import SWR from client components, both ways will allow you to use the SWR client data fetching hooks.
```tsx filename="app/page.tsx" highlight={1} copy
'use client'
import useSWR from 'swr'
export default function Page() {
const { data } = useSWR('/api/user', fetcher)
return
{data.name}
}
```
### Prefetch Data in Server Components
The recommended approach in React Server Components (RSC) is to start fetching with `preload` and pass the returned data to the client component tree through the `cacheData` option of ``:
```tsx filename="app/layout.tsx" copy
import { preload, SWRConfig } from 'swr'
export default async function Layout({ children }: { children: React.ReactNode }) {
const cacheData = {
...preload('/api/user', fetchUserFromAPI),
...preload('/api/posts', fetchPostsFromAPI),
}
return (
{children}
)
}
```
Both `preload` calls start fetching immediately, so the requests run in parallel without being awaited in the layout.
In React Server Components, the promises inside `cacheData` can cross the `"use client"` boundary, and SWR resolves them automatically during Server-Side Rendering:
```tsx filename="app/page.tsx" copy
'use client'
import useSWR from 'swr'
export default function Page() {
// SWR resolves the data preloaded by the Server Component.
// Both `user` and `posts` are ready during SSR and client hydration.
const { data: user } = useSWR('/api/user', fetcher)
const { data: posts } = useSWR('/api/posts', fetcher)
return (
{user.name}'s Posts
{posts.map(post => (
{post.title}
))}
)
}
```
SWR uses the server-loaded result for the initial render. On the client, it takes over and continues with its usual revalidation behavior.
With `preload` and `cacheData`, fetching starts as early as possible on the server. Only the UI boundaries that consume the data, such as the nearest `Suspense` boundary or Next.js layout, are blocked during streaming SSR.
To incrementally adopt this prefetch pattern in your application, you can enable the `strictServerPrefetchWarning` option. This will show a warning message in the console when a key has no pre-filled data provided, helping you identify which data fetching calls could benefit from server-side prefetching.
### How `cacheData` Works
This feature requires SWR 2.5.0-beta.1 or later and is currently experimental.
In a Server Component, `preload(key, fetcher)` starts the fetch immediately and returns a request-scoped `CacheData` object. Pass that object to the `cacheData` option of `SWRConfig`:
```tsx filename="app/page.tsx" copy
import { preload, SWRConfig } from 'swr'
import { User } from './user'
const getUser = () => fetchUserFromDatabase()
export default function Page() {
const cacheData = preload('/api/user', getUser)
return (
)
}
```
The client component uses the same key with its usual client-side fetcher:
```tsx filename="app/user.tsx" copy
'use client'
import useSWR from 'swr'
const fetcher = (url: string) => fetch(url).then(res => res.json())
export function User() {
const { data } = useSWR('/api/user', fetcher)
return
{data?.name}
}
```
SWR uses the server-loaded result for the initial render, writes it to the client cache during hydration, and skips the duplicate initial client request. Later revalidations still use the client-side fetcher.
Each `preload` call returns a new object. Merge the objects to preload multiple keys without awaiting them:
```tsx
const cacheData = {
...preload('/api/user', getUser),
...preload('/api/posts', getPosts),
}
```
`preload` serializes complex SWR keys automatically. The `cacheData` option is only supported by `SWRConfig`; it cannot be passed directly to `useSWR`.
## Client Side Data Fetching
If your page contains frequently updating data, and you don’t need to pre-render the data, SWR is a perfect fit and no special setup is needed: just import `useSWR` and use the hook inside any components that use the data.
Here’s how it works:
* First, immediately show the page without data. You can show loading states for missing data.
* Then, fetch the data on the client side and display it when ready.
This approach works well for user dashboard pages, for example. Because a dashboard is a private, user-specific page, SEO is not relevant and the page doesn’t need to be pre-rendered. The data is frequently updated, which requires request-time data fetching.
## Pre-rendering with Default Data
If the page must be pre-rendered, Next.js supports [2 forms of pre-rendering](https://nextjs.org/docs/basic-features/data-fetching):
**Static Generation (SSG)** and **Server-side Rendering (SSR)**.
Together with SWR, you can pre-render the page for SEO, and also have features such as caching, revalidation, focus tracking, refetching on interval on the client side.
You can use the `cacheData` option of [`SWRConfig`](/docs/global-configuration) to pass server-fetched data into the cache for all SWR hooks inside the boundary.
For example with `getStaticProps`:
```jsx
export async function getStaticProps () {
// `getStaticProps` is executed on the server side.
const article = await getArticleFromAPI()
return {
props: {
cacheData: {
'/api/article': article
}
}
}
}
function Article() {
// `data` is available from the server-provided `cacheData`.
const { data } = useSWR('/api/article', fetcher)
return
{data.title}
}
export default function Page({ cacheData }) {
// SWR hooks inside the `SWRConfig` boundary will use those values.
return (
)
}
```
The page is still pre-rendered. It's SEO friendly, fast to response, but also fully powered by SWR on the client side. The data can be dynamic and self-updated over time.
The `Article` component renders the pre-generated data and hydrates it into the client cache without a duplicate initial request. Later revalidation events still use the client fetcher.
### Complex Keys
`useSWR` can be used with keys that are `array` and `function` types. When creating `cacheData` manually, serialize these keys with `unstable_serialize`.
```jsx
import useSWR, { unstable_serialize } from 'swr'
export async function getStaticProps () {
const article = await getArticleFromAPI(1)
return {
props: {
cacheData: {
// unstable_serialize() array style key
[unstable_serialize(['api', 'article', 1])]: article,
}
}
}
}
function Article() {
// using an array style key.
const { data } = useSWR(['api', 'article', 1], fetcher)
return
{data.title}
}
export default function Page({ cacheData }) {
return (
)
}
```
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Cache
type: conceptual
summary: Customize the cache provider and implement advanced caching strategies.
prerequisites:
- /docs/global-configuration
related:
- /docs/global-configuration
---
# Cache
Upgrade to the latest version (≥ 1.0.0) to use this feature.
In most cases, you shouldn't directly *write* to the cache, which might cause undefined behaviors of SWR. If you need to manually mutate a key, please consider using the SWR APIs.
See also: [Mutation](/docs/mutation), [Reset Cache Between Test Cases](#reset-cache-between-test-cases).
By default, SWR uses a global cache to store and share data across all components. But you can also customize this behavior with the `provider` option of `SWRConfig`.
Cache providers are intended to enable SWR with more customized storages.
## Cache Provider
A cache provider is Map-like object which matches the following TypeScript definition (which can be imported from `swr`):
```typescript
interface Cache {
get(key: string): Data | undefined
set(key: string, value: Data): void
delete(key: string): void
keys(): IterableIterator
}
```
For example, a [JavaScript Map](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map) instance can be directly used as the cache provider for SWR.
## Create Cache Provider
The `provider` option of `SWRConfig` receives a function that returns a [cache provider](#cache-provider). The provider will then be used by all SWR hooks inside that `SWRConfig` boundary. For example:
```jsx
import useSWR, { SWRConfig } from 'swr'
function App() {
return (
new Map() }}>
)
}
```
All SWR hooks inside `` will read and write from that Map instance. You can also use other cache provider implementations as well for your specific use case.
In the example above, when the `` component is re-mounted, the provider will also be re-created. Cache providers should be put higher in the component tree, or outside of render.
When nested, SWR hooks will use the upper-level cache provider. If there is no upper-level cache provider, it fallbacks to the default cache provider, which is an empty `Map`.
If a cache provider is used, the global `mutate` will **not** work for SWR hooks under that `` boundary. Please use [this](#access-current-cache-provider) instead.
## Access Current Cache Provider
When inside a React component, you need to use the [`useSWRConfig`](/docs/global-configuration#access-to-global-configurations) hook to get access to the current cache provider as well as other configurations including `mutate`:
```jsx
import { useSWRConfig } from 'swr'
function Avatar() {
const { cache, mutate, ...extraConfig } = useSWRConfig()
// ...
}
```
If it's not under any ``, it will return the default configurations.
## Experimental: Extend Cache Provider
This is an experimental feature, the behavior might change in future upgrades.
When multiple `` components are nested, cache provider can be extended.
The first argument for the `provider` function is the cache provider of the upper-level `` (or the default cache if there's no parent ``), you can use it to extend the cache provider:
```jsx
newCache }}>
...
```
## Examples
### LocalStorage Based Persistent Cache
You might want to sync your cache to `localStorage`. Here's an example implementation:
```jsx
function localStorageProvider() {
// When initializing, we restore the data from `localStorage` into a map.
const map = new Map(JSON.parse(localStorage.getItem('app-cache') || '[]'))
// Before unloading the app, we write back all the data into `localStorage`.
window.addEventListener('beforeunload', () => {
const appCache = JSON.stringify(Array.from(map.entries()))
localStorage.setItem('app-cache', appCache)
})
// We still use the map for write & read for performance.
return map
}
```
Then use it as a provider:
```jsx
```
As an improvement, you can also use the memory cache as a buffer, and write to `localStorage` periodically. You can also implement a similar layered cache with IndexedDB or WebSQL.
### Reset Cache Between Test Cases
When testing your application, you might want to reset the SWR cache between test cases. You can simply wrap your application with an empty cache provider. Here's an example with Jest:
```jsx
describe('test suite', async () => {
it('test case', async () => {
render(
new Map() }}>
)
})
})
```
### Modify the Cache Data
You should not write to the cache directly, it might cause undefined behavior.
You can use [`mutate`](/docs/mutation) to modify the cache. For example, you can clear all cache data like the following.
```jsx
const { mutate } = useSWRConfig()
mutate(
key => true, // which cache keys are updated
undefined, // update cache data to `undefined`
{ revalidate: false } // do not revalidate
)
```
More information can be found [here](/docs/arguments#multiple-arguments).
## Clear the Cache
This API requires SWR 2.5.0-beta.1 or later.
`unload()` clears every entry in a cache provider and notifies all mounted hooks in that provider scope. It also clears pending preloads and makes SWR ignore the results of requests and mutations that were already in flight. The underlying network operations are not aborted.
By default, mounted hooks revalidate after the cache is cleared:
```tsx
import { unload } from 'swr'
unload()
```
To clear the cache without immediately refetching, disable revalidation:
```tsx
unload({ revalidate: false })
```
Those hooks re-render with `data` set to `undefined` and remain empty until another event revalidates them, such as focus, reconnect, or remount. `unload` also drops data retained by `keepPreviousData` and clears internal keys used by APIs such as `useSWRInfinite`, which a `mutate` key filter might not match.
The top-level import is bound to SWR's default cache provider. For a custom provider, get the scoped `unload` function from `useSWRConfig`:
```tsx
import { useSWRConfig } from 'swr'
function ClearCacheButton() {
const { unload } = useSWRConfig()
return (
)
}
```
### API
```ts
unload(options?: { revalidate?: boolean }): void
```
* `revalidate = true`: revalidate the keys used by mounted hooks after clearing the cache.
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: DevTools
type: guide
summary: Debug SWR with developer tools.
prerequisites:
- /docs/getting-started
related:
- /docs/advanced/cache
---
# DevTools
SWRDevTools is not an official project of Vercel.
[SWRDevTools](https://swr-devtools.vercel.app/) is a developer tool for SWR, which helps to debug your SWR cache and fetchers.
You can install SWR DevTools from the extension pages and use it with zero settings!
* Chrome: [https://chrome.google.com/webstore/detail/swr-devtools/liidbicegefhheghhjbomajjaehnjned](https://chrome.google.com/webstore/detail/swr-devtools/liidbicegefhheghhjbomajjaehnjned)
* Firefox: [https://addons.mozilla.org/en-US/firefox/addon/swr-devtools/](https://addons.mozilla.org/en-US/firefox/addon/swr-devtools/)
After installing it, the SWR devtool panel will appear on browsers' developer tools.
Checkout more information on the [website](https://swr-devtools.vercel.app/) and the [repository](https://github.com/koba04/swr-devtools)
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Performance
type: conceptual
summary: Optimize performance with deduplication, deep comparison, and dependency tracking.
prerequisites:
- /docs/getting-started
related:
- /docs/data-fetching
- /docs/advanced/cache
---
# Performance
SWR provides critical functionality in all kinds of web apps, so **performance** is a top priority.
SWR’s built-in **caching** and **[deduplication](#deduplication)** skips unnecessary network requests, but
the performance of the `useSWR` hook itself still matters. In a complex app, there could be hundreds of `useSWR` calls in a single page render.
SWR ensures that your app has:
* *no unnecessary requests*
* *no unnecessary re-renders*
* *no unnecessary code imported*
without any code changes from you.
## Deduplication
It’s very common to reuse SWR hooks in your app. For example, an app that renders the current user’s avatar 5 times:
```jsx
function useUser () {
return useSWR('/api/user', fetcher)
}
function Avatar () {
const { data, error } = useUser()
if (error) return
if (!data) return
return
}
function App () {
return <>
}
```
Each `` component has a `useSWR` hook inside. Since they have the same SWR key and are rendered almost at the same time, **only 1 network request will be made**.
You can reuse your data hooks (like `useUser` in the example above) everywhere, without worrying about performance or duplicated requests.
There is also a [`dedupingInterval` option](/docs/api) for overriding the default deduplication interval.
## Deep Comparison
SWR **deep compares** data changes by default. If the `data` value isn’t changed, a re-render will not be triggered.
You can also customize the comparison function via the [`compare` option](/docs/api) if you want to change the behavior.
For example, some API responses return a server timestamp that you might want to exclude from the data diff.
## Dependency Collection
`useSWR` returns 4 **stateful** values: `data`, `error`, `isLoading` and `isValidating`, each one can be updated independently.
For example, if we print those values within a full data-fetching lifecycle, it will be something like this:
```jsx
function App () {
const { data, error, isLoading, isValidating } = useSWR('/api', fetcher)
console.log(data, error, isLoading, isValidating)
return null
}
```
In the worst case (the first request failed, then the retry was successful), you will see 4 lines of logs:
```js
// console.log(data, error, isLoading, isValidating)
undefined undefined true true // => start fetching
undefined Error false false // => end fetching, got an error
undefined Error true true // => start retrying
Data undefined false false // => end retrying, get the data
```
The state changes make sense. But that also means our component **rendered 4 times**.
If we change our component to only use `data`:
```jsx
function App () {
const { data } = useSWR('/api', fetcher)
console.log(data)
return null
}
```
The magic happens — there are only **2 re-renders** now:
```js
// console.log(data)
undefined // => hydration / initial render
Data // => end retrying, get the data
```
The exact same process has happened internally, there was an error from the first request, then we got the data from the retry.
However, **SWR only updates the states that are used by the component**, which is only `data` now.
If you are not always using all these 3 states, you are already benefitting from this feature.
At [Vercel](https://vercel.com), this optimization results in \~60% fewer re-renders.
## Tree Shaking
The SWR package is [tree-shakeable](https://webpack.js.org/guides/tree-shaking) and side-effect free.
That means if you are only importing the core `useSWR` API, unused APIs like `useSWRInfinite` won't be bundled in your application.
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: React Native
type: integration
summary: Use SWR with React Native applications.
prerequisites:
- /docs/getting-started
related:
- /docs/getting-started
---
# React Native
Upgrade to the latest version (≥ 1.0.0) to experience this customization.
Unlike React running inside the browsers, React Native has a very different user experience. For example there is no “tab focus”, switching from the background to the app is considered as a “focus” instead.
To customize these behaviors, you can replace the default browser `focus` and `online` events listeners with React Native’s app state detection and other native ported API, and configure SWR to use them.
## Example
### Global Setup
You can wrap your app under `SWRConfig` and preconfig all configurations there
```jsx
```
### Customize `focus` and `reconnect` Events
There're few configurations you need to take care of such as `isOnline`, `isVisible`, `initFocus` and `initReconnect`.
`isOnline` and `isVisible` are functions that return a boolean, to determine if the application is "active". By default, SWR will bail out a revalidation if these conditions are not met.
When using `initFocus` and `initReconnect`, it's required to also set up a [custom cache provider](/docs/advanced/cache). You can use an empty `Map()` or any storage you prefer.
```jsx
new Map(),
isOnline() {
/* Customize the network state detector */
return true
},
isVisible() {
/* Customize the visibility state detector */
return true
},
initFocus(callback) {
/* Register the listener with your state provider */
},
initReconnect(callback) {
/* Register the listener with your state provider */
}
}}
>
```
Let's take `initFocus` as example:
```jsx
import { AppState } from 'react-native'
// ...
new Map(),
isVisible: () => { return true },
initFocus(callback) {
let appState = AppState.currentState
const onAppStateChange = (nextAppState) => {
/* If it's resuming from background or inactive mode to active one */
if (appState.match(/inactive|background/) && nextAppState === 'active') {
callback()
}
appState = nextAppState
}
// Subscribe to the app state change events
const subscription = AppState.addEventListener('change', onAppStateChange)
return () => {
subscription.remove()
}
}
}}
>
```
For `initReconnect`, it requires some 3rd party libraries such as [NetInfo](https://github.com/react-native-netinfo/react-native-netinfo) to subscribe to the network status. The implementation will be similar to the example above: receiving a `callback` function and trigger it when the network recovers from offline, so SWR can start a revalidation to keep your data up-to-date.
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Understanding SWR
type: conceptual
summary: Understand how SWR works internally.
prerequisites:
- /docs/getting-started
related:
- /docs/advanced/cache
- /docs/advanced/performance
---
# Understanding SWR
## State Machine
`useSWR` returns `data`, `error`, `isLoading`, and `isValidating` depending on the state of the `fetcher` function. This diagrams describe how SWR returns values in some scenarios.
### Fetch and Revalidate
This pattern is to fetch data and revalidate it later.
### Key Change
This pattern is to fetch data and change the key and revalidate it later.
### Key Change + Previous Data
This pattern is to fetch data and change the key and revalidate it later with the `keepPreviousData` option.
### Fallback
This pattern is to fetch data and revalidate it later with fallback data.
### Key Change + Fallback
This pattern is to fetch data and change the key and revalidate it later with fallback data.
### Key Change + Previous Data + Fallback
This pattern is to fetch data and change the key and revalidate it later with the `keepPreviousData` option and fallback data.
## Combining with isLoading and isValidating for better UX
Comparing to the existing `isValidating` value, `isLoading` is a new property that can help you for the more general loading cases for UX.
* `isValidating` becomes `true` whenever there is an ongoing request **whether the data is loaded or not**
* `isLoading` becomes `true` when there is an ongoing request and **data is not loaded yet**.
Simply saying you can use `isValidating` for indicating everytime there is an ongoing revalidation, and `isLoading` for indicating that SWR is revalidating but there is no data yet to display.
Fallback data and previous data are not considered "loaded data," so when you use fallback data or enable the keepPreviousData option, you might have data to display.
```jsx
function Stock() {
const { data, isLoading, isValidating } = useSWR(STOCK_API, fetcher, {
refreshInterval: 3000
});
// If it's still loading the initial data, there is nothing to display.
// We return a skeleton here.
if (isLoading) return ;
// Otherwise, display the data and a spinner that indicates a background
// revalidation.
return (
<>
${data}
{isValidating ? : null}
);
}
```
You can find the code example [here](https://codesandbox.io/s/swr-isloading-jtopow)
## Return previous data for better UX
When doing data fetching based on continuous user actions, e.g. real-time search when typing, keeping the previous fetched data can improve the UX a lot. `keepPreviousData` is an option to enable that behavior. Here's a simple search UI:
```jsx
function Search() {
const [search, setSearch] = React.useState('');
const { data, isLoading } = useSWR(`/search?q=${search}`, fetcher, {
keepPreviousData: true
});
return (
);
}
```
With `keepPreviousData` enabled, you will still get the previous data even if you change the SWR key and the data for the new key starts loading again.
You can find the full code for this example here: [https://codesandbox.io/s/swr-keeppreviousdata-fsjz3m](https://codesandbox.io/s/swr-keeppreviousdata-fsjz3m).
## Dependency Collection for performance
SWR only triggers re-rendering when the states used in the component have been updated. If you only use `data` in the component, SWR ignores the updates of other properties like `isValidating`, and `isLoading`. This reduces rendering counts a lot. More information can be found [here](/docs/advanced/performance#dependency-collection).
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Announcing SWR 1.0
description: Almost 2 years ago we open sourced SWR, the tiny data-fetching React library that people love. Today we are reaching another milestone: the 1.0 version of SWR.
type: overview
summary: Announcing SWR 1.0 with new features and improvements.
---
# Announcing SWR 1.0
Almost 2 years ago we [open sourced](https://twitter.com/vercel/status/1188911002626097157) SWR, the tiny data-fetching React library that people love. Today we are reaching another milestone: the 1.0 version of SWR!
## What’s New
### Smaller Size
[Performance](/docs/advanced/performance) is one of the most important features of SWR. In 1.0, we made the library significantly smaller **without removing any existing features**:
* 41% smaller core (24% smaller when gzipped, **3.9 kB**)
* 52% smaller package install size
* Improved tree-shaking
There are many reasons to make the library lightweight: your application will have a smaller bundle, a leaner runtime, and a smaller `node_modules` directory.
We’ve also improved the bundling of the package, and it now supports path imports:
```js
import useSWR from 'swr'
import useSWRInfinite from 'swr/infinite'
```
If you are not using `useSWRInfinite`, it will not be included in your application.
### Fallback Data
In 1.0, there’s a new `fallback` option that you can provide any pre-fetched data as the initial value of all SWR hooks with specific keys:
```jsx
```
This is very helpful for scenarios such as SSG, SSR, and data mockup for testing. Check the docs [Next.js SSG and SSR](/docs/with-nextjs) for more details.
For better consistency and to avoid confusion, the old `initialData` is now renamed to `fallbackData`, which still provides a single fallback value for the given hook.
### Immutable Mode
Sometimes you want to mark a resource as **immutable** if it will never change. It's better to disable automatic revalidations for it and only make the request once. There is now a helper hook to make this easier:
```jsx
import useSWRImmutable from 'swr/immutable'
// ...
useSWRImmutable(key, fetcher, options)
```
It has the exact same API as the `useSWR` hook, but it will never revalidate upon tab focus or network recovery. There's also a new option, `revalidateIfStale`, you can use to control the behavior precisely. More information can be found [here](/docs/revalidation#disable-automatic-revalidations).
### Custom Cache Provider
By default, SWR uses a single global cache to store all the data. In 1.0, you are able to customize it with the new `provider` option:
```jsx
myCache
}}>
```
You can use this new feature to do many powerful things. We have a couple of examples here: [*Mutate Multiple Keys with RegEx*](/docs/advanced/cache#mutate-multiple-keys-from-regex), [*LocalStorage Based Persistent Cache*](/docs/advanced/cache#localstorage-based-persistent-cache), [*Reset Cache Between Tests*](/docs/advanced/cache#reset-cache-between-test-cases).
This new cache provider API is also more compatible with concurrent rendering of React 18. If you are adding a cache provider, make sure to use the global `mutate` function returned from `useSWRConfig()`.
You can read the docs [Cache Provider](/docs/advanced/cache) for more details.
### useSWRConfig()
There is a new Hook API to return all global configurations, including the current cache provider and global `mutate` function:
```jsx
import { useSWRConfig } from 'swr'
function Foo () {
const { refreshInterval, cache, mutate, ...restConfig } = useSWRConfig()
// ...
}
```
More information can be found [here](/docs/global-configuration#access-to-global-configurations).
### Middleware
SWR Middleware provide a new way for you to build and reuse abstractions on top of SWR hooks:
```jsx
// ... or directly in `useSWR`:
useSWR(key, fetcher, { use: [...middleware] })
```
A lot of new ideas can be implemented with this feature, and we've built some examples: [*Request Logger*](/docs/middleware#request-logger), [*Keep Previous Data When Changing the Key*](/docs/middleware#keep-previous-result), and [*Serialize Object Keys*](/docs/middleware#serialize-object-keys).
Check the [Middleware API](/docs/middleware) for more details.
### Improvements and Better Test Coverage
Since 0.x, we've made hundreds of small improvements and bugfixes. SWR now has 157 tests that cover most of the edge cases in data fetching. Read the [Changelog](https://github.com/vercel/swr/releases) for more details.
### Docs Translations
Thanks to our [contributors](https://github.com/vercel/swr-site/graphs/contributors) and Nextra’s [i18n feature](https://nextra.vercel.app/features/i18n), we now offer SWR documentation in six different languages:
* [English](https://swr.vercel.app)
* [Spanish](https://swr.vercel.app/es-ES)
* [Simplified Chinese](https://swr.vercel.app/zh-CN)
* [Japanese](https://swr.vercel.app/ja)
* [Korean](https://swr.vercel.app/ko)
* [Russian](https://swr.vercel.app/ru)
## Migration Guide
### Update `useSWRInfinite` Imports
`useSWRInfinite` needs to be imported from `swr/infinite`:
```diff
- import { useSWRInfinite } from 'swr'
+ import useSWRInfinite from 'swr/infinite'
```
If you are using the corresponding types, update the import path too:
```diff
- import { SWRInfiniteConfiguration, SWRInfiniteResponse } from 'swr'
+ import { SWRInfiniteConfiguration, SWRInfiniteResponse } from 'swr/infinite'
```
### Change `revalidate` to `mutate`
`useSWR` no longer returns the `revalidate` method, change to `mutate` instead:
```diff
- const { revalidate } = useSWR(key, fetcher, options)
+ const { mutate } = useSWR(key, fetcher, options)
// ...
- revalidate()
+ mutate()
```
### Rename `initialData` to `fallbackData`
```diff
- useSWR(key, fetcher, { initialData: ... })
+ useSWR(key, fetcher, { fallbackData: ... })
```
### No More Default Fetcher
SWR no longer provides the default fetcher (a `fetch` call that parses the data as JSON). The easiest way to migrate the change is to use the `` component:
```jsx
fetch(url).then(res => res.json()) }}>
// ... or
useSWR(key, (url) => fetch(url).then(res => res.json()))
```
### Recommend to Use the Hook-Returned `mutate`
This is **not** a breaking change, but we will now *recommend* to always use the `mutate` returned from the `useSWRConfig` hook:
```diff
- import { mutate } from 'swr'
+ import { useSWRConfig } from 'swr'
function Foo () {
+ const { mutate } = useSWRConfig()
return
}
```
If you are not using a cache provider, the current global import `import { mutate } from 'swr'` still works.
### Renamed Types
If you are using TypeScript, the following type names have been changed for consistency:
| 0.x (deprecated) | 1.0 | Note |
| ------------------------------ | -------------------------- | ----------------------- |
| `ConfigInterface` | `SWRConfiguration` | |
| `keyInterface` | `Key` | |
| `responseInterface` | `SWRResponse` | |
| `RevalidateOptionInterface` | `RevalidatorOptions` | |
| `revalidateType` | `Revalidator` | |
| `SWRInfiniteResponseInterface` | `SWRInfiniteResponse` | moved to `swr/infinite` |
| `SWRInfiniteConfigInterface` | `SWRInfiniteConfiguration` | moved to `swr/infinite` |
### Beta and Unofficial Feature Users
If you are using a beta version of SWR, or using any undocumented APIs, please be aware of the following changes:
* `import { cache } from 'swr'` is removed; use the new [`useSWRConfig` API](#useswrconfig) instead.
* `import { createCache } from 'swr'` is removed; use the new [Cache Provider API](/docs/advanced/cache) instead.
* `revalidateWhenStale` is renamed to `revalidateIfStale`.
* `middlewares` is renamed to `use`.
### Changelog
Read the full Changelog [on GitHub](https://github.com/vercel/swr/releases).
## What’s Next
In future releases, we will keep improving the library while maintaining the stability. We are also aiming to embrace future React versions, as several new features and improvements in 1.0 are already preparing for that. In addition, we are also working on new features to improve the experience of doing data fetching in React and the experience of using this library.
If you have any feedback about this release, please [let us know](https://github.com/vercel/swr/discussions).
## Thank You!
Special thanks to [Toru Kobayashi](https://twitter.com/koba04) and [Yixuan Xu](https://twitter.com/yixuanxu94) for their contributions to the library, and [Paco Coursey](https://twitter.com/pacocoursey), [uttk](https://github.com/uttk), [Tomohiro SHIOYA](https://github.com/shioyang), [Markoz Peña](https://github.com/markozxuu), [SeulGi Choi](https://github.com/cs09g), [Fang Lu](https://github.com/huzhengen), [Valentin Politov](https://github.com/valentinpolitov) for their work on the translations and docs. This release can't happen without them.
We also want to thank the entire community, our [110 contributors](https://github.com/vercel/swr/graphs/contributors) (+ [45 docs contributors](https://github.com/vercel/swr-site/graphs/contributors)), and everyone who helped and gave us feedback!
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Announcing SWR 2.0
description: Announcing SWR 2.0: new mutation APIs and improved optimistic UI capabilities, new DevTools, better support for concurrent rendering, and more.
type: overview
summary: Announcing SWR 2.0 with new mutation APIs, DevTools, and optimistic UI.
---
# Announcing SWR 2.0
We are thrilled to announce the release of SWR 2.0, the popular React data-fetching library that enables components to fetch, cache, and mutate data and keeps the UI up-to-date with changes in that data over time.
This new version comes packed with improvements and new features, such as new mutation APIs, improved optimistic UI capabilities, new DevTools, and better support for concurrent rendering. We would like to extend a huge thank you to all the contributors and maintainers who made this release possible.
## Mutation and Optimistic UI
### useSWRMutation
Mutation is an important part of the data-fetching process. They allow you to make changes to your data both locally and remotely. Our existing `mutate` API allows you to revalidate and mutate resources manually. In SWR 2.0, the new hook `useSWRMutation` makes it even simpler to remotely change data using a declarative API. You can set up a mutation using the hook, and then activate it later:
```jsx {11,16}
import useSWRMutation from 'swr/mutation'
async function sendRequest(url, { arg }) {
return fetch(url, {
method: 'POST',
body: JSON.stringify(arg)
})
}
function App() {
const { trigger, isMutating } = useSWRMutation('/api/user', sendRequest)
return (
)
}
```
The example above defines a `sendRequest` mutation that affects the `'/api/user'` resource. Unlike `useSWR`, `useSWRMutation` will not immediately start the request upon rendering. Instead, it returns a `trigger` function that can later be called to manually start the mutation.
The `sendRequest` function will be called when the button is clicked, with the extra argument `{ username: 'johndoe' }`. The value of `isMutating` will be set to `true` until the mutation has finished.
Additionally, this new hook addresses other issues you may have with mutations:
* Optimistically update the UI while data is being mutated
* Automatically revert when mutation fails
* Avoid any potential race conditions between `useSWR` and other mutations of the same resource
* Populate the `useSWR` cache after mutation completes
* ...
You can find in-depth API references and examples by reading the [docs](/docs/mutation#useswrmutation) or scrolling through the next few sections.
### Optimistic UI
Optimistic UI is an excellent model for creating websites that feel fast and responsive; however, it can be difficult to implement correctly. SWR 2.0 has added some new powerful options to make it easier.
Let’s say we have an API that adds a new todo to the todo list and sends it to the server:
```jsx
await addNewTodo('New Item')
```
In our UI, we use a `useSWR` hook to display the todo list, with an “Add New Item” button that triggers this request and asks SWR to re-fetch the data via `mutate()`:
```jsx {7,8}
const { mutate, data } = useSWR('/api/todos')
return <>
{/* Display data */}
```
However, the `await addNewTodo(...)` request could be very slow. When it’s ongoing, users still see the old list even if we can already know what the new list will look like. With the new `optimisticData` option, we can show the new list optimistically, before the server responds:
```jsx {8}
const { mutate, data } = useSWR('/api/todos')
return <>
{/* Display data */}
```
SWR will immediately update the `data` with the `optimisticData` value, and then send the request to the server. Once the request finishes, SWR will revalidate the resource to ensure it’s the latest.
Like many APIs, if the `addNewTodo(...)` request returns us the latest data from the server, we can directly show that result, too (instead of starting a new revalidation)! There’s the new `populateCache` option to tell SWR to update the local data with the mutate response:
```jsx {9}
const { mutate, data } = useSWR('/api/todos')
return <>
{/* Display data */}
```
At the same time, we don’t need another revalidation afterward as the response data is from the source of truth, we can disable it with the `revalidate` option:
```jsx {10}
const { mutate, data } = useSWR('/api/todos')
return <>
{/* Display data */}
```
Lastly, if `addNewTodo(...)` fails with an exception, we can revert the optimistic data (`[...data, 'New Item']`) we just set, by setting `rollbackOnError` to `true` (which is also the default option). When that happens, SWR will roll back `data` to the previous value.
```jsx {11}
const { mutate, data } = useSWR('/api/todos')
return <>
{/* Display data */}
```
All these APIs are supported in the new `useSWRMutation` hook as well. To learn more about them, you can check out our [docs](/docs/mutation#optimistic-updates). And here is a demo showing that behavior:
### Mutate Multiple Keys
The global `mutate` API now accepts a filter function, where you can mutate or revalidate specific keys. This will be helpful for use cases such as invalidating all the cached data. To learn more, you can read [Mutate Multiple Keys](/docs/mutation#mutate-multiple-items) in the docs.
```jsx
import { mutate } from 'swr'
// Or from the hook if you have customized your cache provider:
// { mutate } = useSWRConfig()
// Mutate single resource
mutate(key)
// Mutate multiple resources and clear the cache (set to undefined)
mutate(
key => typeof key === 'string' && key.startsWith('/api/item?id='),
undefined,
{ revalidate: false }
)
```
## SWR DevTools
[SWRDevTools](https://swr-devtools.vercel.app) is a browser extension that helps you debug your SWR cache and the fetch results. Check our [devtools](/docs/advanced/devtools) section for how to use devtools in your application.
## Preloading Data
Preloading data can improve the user experience tremendously. If you know the resource is going to be used later in the application, you can use the new `preload` API to start fetching it early:
```jsx {6}
import useSWR, { preload } from 'swr'
const fetcher = (url) => fetch(url).then((res) => res.json())
// You can call the preload function in anywhere
preload('/api/user', fetcher)
function Profile() {
// The component that actually uses the data:
const { data, error } = useSWR('/api/user', fetcher)
// ...
}
export function Page () {
return
}
```
In this example, the `preload` API is called in the global scope. This means that we start to preload the resource before React even starts to render anything.
And when the `Profile` component is being rendered, the data can probably be available already. If it’s still ongoing, the `useSWR` hook will reuse that ongoing preloading request instead of starting a new one.
The `preload` API can also be used in cases like preloading data for another page that will likely be rendered. More information about prefetching data with SWR can be found [here](/docs/prefetching).
## `isLoading`
`isLoading` is a new state returned by `useSWR`, that indicates **if the request is still ongoing, and there is no data loaded yet**. Previously, the `isValidating` state represents both the initial loading state and revalidating state so we had to check if both `data` and `error` are `undefined` to determine if it was the initial loading state.
Now, it is so easy that you can directly use the `isLoading` value to render a loading message:
```jsx
import useSWR from 'swr'
function Profile() {
const { data, isLoading } = useSWR('/api/user', fetcher)
if (isLoading) return
loading...
return
hello {data.name}!
}
```
Note that `isValidating` is still present so you can still use it to show a loading indicator for revalidations.
We have added the new [Understanding SWR](/docs/advanced/understanding) page to describe how SWR returns values, which includes the difference between `isValidating` and `isLoading`, and how to combine them to improve user experience.
## Preserving Previous State
The `keepPreviousData` option is a new addition that allows you to keep the data that was fetched before. This improves UX immensely when you’re fetching data based on user actions happening in real time, like with a live search feature, where the resource’s `key` keeps changing:
```jsx {5}
function Search() {
const [search, setSearch] = React.useState('');
const { data, isLoading } = useSWR(`/search?q=${search}`, fetcher, {
keepPreviousData: true
})
return (
);
}
```
Check the code on [CodeSandbox](https://codesandbox.io/s/swr-keeppreviousdata-fsjz3m) and you can read more about it [here](/docs/advanced/understanding#return-previous-data-for-better-ux).
## Extending Configurations
`SWRConfig` can now accept a function value. When you have multiple levels of ``, the inner receives the parent configuration and returns a new one. This change makes it more flexible to configure SWR in a large codebase. More information can be found [here](/docs/global-configuration).
```jsx
({
dedupingInterval: parentConfig.dedupingInterval * 5,
refreshInterval: 100,
})}
>
```
## Improved React 18 Support
SWR has updated its internal code to use `useSyncExternalStore` and `startTransition` APIs in React 18. These ensure stronger consistency when rendering UI concurrently. This change doesn’t require any user code changes and all developers will benefit from it directly. Shims are included for React 17 and below.
SWR 2.0 and all the new features are still compatible with React 16 and 17.
## Migration Guide
### Fetcher No Longer Accepts Multiple Arguments
`key` is now passed as a single argument.
```diff
- useSWR([1, 2, 3], (a, b, c) => {
+ useSWR([1, 2, 3], ([a, b, c]) => {
assert(a === 1)
assert(b === 2)
assert(c === 3)
})
```
### Global Mutate No Longer Accepts a `getKey` Function
Now, if you pass a function to the global `mutate`, it will be used as a [filter](/blog/swr-v2#mutate-multiple-keys). Previously, you can pass a function that returns a key to the global `mutate`:
```diff
- mutate(() => '/api/item') // a function to return a key
+ mutate('/api/item') // to mutate the key, directly pass it
```
### New Required Property `keys()` for Cache Interface
When you use your own cache implementation, the Cache interface now requires a `keys()` method that returns all keys in the cache object, similar to the JavaScript Map instances.
```diff
interface Cache {
get(key: string): Data | undefined
set(key: string, value: Data): void
delete(key: string): void
+ keys(): IterableIterator
}
```
### Changed Cache Internal Structure
The internal structure of the cache data will be an object that holds all the current states.
```diff
- assert(cache.get(key) === data)
+ assert(cache.get(key) === { data, error, isValidating })
// getter
- cache.get(key)
+ cache.get(key)?.data
// setter
- cache.set(key, data)
+ cache.set(key, { ...cache.get(key), data })
```
You should not write to the cache directly, it might cause undefined behavior.
### `SWRConfig.default` Is Renamed as `SWRConfig.defaultValue`
`SWRConfig.defaultValue` is the property for accessing the default SWR config.
```diff
- SWRConfig.default
+ SWRConfig.defaultValue
```
### Type `InfiniteFetcher` Is Renamed as `SWRInfiniteFetcher`
```diff
- import type { InfiniteFetcher } from 'swr/infinite'
+ import type { SWRInfiniteFetcher } from 'swr/infinite'
```
### Avoid Suspense on Server
If you want to use `suspense: true` with SWR on the server-side, including pre-rendering in Next.js, then you must provide initial data via [`fallbackData` or `fallback`](/docs/with-nextjs#pre-rendering-with-default-data). Today, this means that you can't use Suspense to fetch data on the server side. Your other two options are doing fully client-side data-fetching or getting your framework to fetch the data for you (like getStaticProps does in Next.js).
### ES2018 as the Build Target
If you want to support IE 11, you have to target ES5 in your framework or a bundler. This change has made a performance improvement on SSR, and keeps the bundle size small.
## Changelog
Read the full Changelog [on GitHub](https://github.com/vercel/swr/releases).
## The Future & Thank You!
With the new release of [Next.js 13](https://nextjs.org/blog/next-13), we see a lot of exciting new things as well as paradigm shifts in the React ecosystem: [React Server Components](https://beta.nextjs.org/docs/rendering/server-and-client-components), streaming SSR, [async components](https://beta.nextjs.org/docs/data-fetching/fetching#asyncawait-in-server-components), and the [`use` hook](https://github.com/acdlite/rfcs/blob/first-class-promises/text/0000-first-class-support-for-promises.md#usepromise). Many of them are related to data-fetching, and some of them have overlapping use cases with SWR.
However, the goal of the SWR project remains the same. We want it to be a drop-in library that is lightweight, framework agnostic, and a little bit *opinionated* (i.e. revalidate upon focus). Instead of trying to be a standard solution, we want to focus on innovations that make the UX better. In the meantime, we are also doing research on how to improve SWR with these new abilities of React.
We want to thank every one of the [143](https://github.com/vercel/swr/graphs/contributors) contributors (+ [106](https://github.com/vercel/swr-site/graphs/contributors) docs contributors), as well as those who helps us out or gave feedback. A special thanks goes to [Toru Kobayashi](https://twitter.com/koba04) for all his work on DevTools and docs– we couldn’t have done it without you!
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Authentication
type: guide
summary: Authentication state management with SWR.
---
# Authentication
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Basic Usage
type: guide
summary: Basic data fetching with SWR.
---
# Basic Usage
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Error Handling
type: guide
summary: Error handling with SWR.
---
# Error Handling
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Infinite Loading
type: guide
summary: Infinite scrolling with useSWRInfinite.
---
# Infinite Loading
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Optimistic UI
type: guide
summary: Optimistic UI updates with SWR mutation.
---
# Optimistic UI
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Next.js SSR
type: guide
summary: Server-side rendering with SWR and Next.js.
---
# Next.js SSR
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
---
title: Subscription
type: guide
summary: Real-time data subscription with SWR.
---
# Subscription
---
For a semantic overview of all documentation, see [/sitemap.md](/sitemap.md)
For an index of all available documentation, see [/llms.txt](/llms.txt)
For agent-facing discovery, including API and MCP surfaces, see [/agents.md](/agents.md)
