Async Rate Limiting Guide

All core concepts from the Rate Limiting Guide apply to async rate limiting as well.

You can usually just use the normal synchronous rate limiter and it will work with async functions, but for advanced use cases, such as wanting to use the return value of a rate-limited function (instead of just calling a setState side effect), or putting your error handling logic in the rate limiter, you can use the async rate limiter.

TanStack Pacer provides async rate limiting through the AsyncRateLimiter class and the asyncRateLimit function.

Here's a basic example showing how to use the async rate limiter for an API operation:

const rateLimitedApi = asyncRateLimit(
  async (id: string) => {
    const response = await fetch(`/api/data/${id}`)
    return response.json()
  },
  {
    limit: 5,
    window: 1000,
    onExecute: (limiter) => {
      console.log('API call succeeded:', limiter.getExecutionCount())
    },
    onReject: (limiter) => {
      console.log(`Rate limit exceeded. Try again in ${limiter.getMsUntilNextWindow()}ms`)
    },
    onError: (error, limiter) => {
      console.error('API call failed:', error)
    }
  }
)

// Usage
try {
  const result = await rateLimitedApi('123')
  // Handle successful result
} catch (error) {
  // Handle errors if no onError handler was provided
  console.error('API call failed:', error)
}

const rateLimitedApi = asyncRateLimit(
  async (id: string) => {
    const response = await fetch(`/api/data/${id}`)
    return response.json()
  },
  {
    limit: 5,
    window: 1000,
    onExecute: (limiter) => {
      console.log('API call succeeded:', limiter.getExecutionCount())
    },
    onReject: (limiter) => {
      console.log(`Rate limit exceeded. Try again in ${limiter.getMsUntilNextWindow()}ms`)
    },
    onError: (error, limiter) => {
      console.error('API call failed:', error)
    }
  }
)

// Usage
try {
  const result = await rateLimitedApi('123')
  // Handle successful result
} catch (error) {
  // Handle errors if no onError handler was provided
  console.error('API call failed:', error)
}

Unlike the synchronous rate limiter which returns a boolean indicating success, the async version allows you to capture and use the return value from your rate-limited function. The maybeExecute method returns a Promise that resolves with the function's return value, allowing you to await the result and handle it appropriately.

The async rate limiter provides robust error handling capabilities:

• If your rate-limited function throws an error and no onError handler is provided, the error will be thrown and propagate up to the caller
• If you provide an onError handler, errors will be caught and passed to the handler instead of being thrown
• The throwOnError option can be used to control error throwing behavior:
  • When true (default if no onError handler), errors will be thrown
  • When false (default if onError handler provided), errors will be swallowed
  • Can be explicitly set to override these defaults
• You can track error counts using getErrorCount() and check execution state with getIsExecuting()
• The rate limiter maintains its state and can continue to be used after an error occurs
• Rate limit rejections (when limit is exceeded) are handled separately from execution errors via the onReject handler

The AsyncRateLimiter supports the following callbacks:

• onSuccess: Called after each successful execution, providing the result and rate limiter instance
• onSettled: Called after each execution (success or failure), providing the rate limiter instance
• onError: Called if the async function throws an error, providing both the error and the rate limiter instance

Both the Async and Synchronous rate limiters support the onReject callback for handling blocked executions.

Example:

const asyncLimiter = new AsyncRateLimiter(async (id) => {
  await saveToAPI(id)
}, {
  limit: 5,
  window: 1000,
  onExecute: (rateLimiter) => {
    // Called after each successful execution
    console.log('Async function executed', rateLimiter.getExecutionCount())
  },
  onReject: (rateLimiter) => {
    // Called when an execution is rejected
    console.log(`Rate limit exceeded. Try again in ${rateLimiter.getMsUntilNextWindow()}ms`)
  },
  onError: (error) => {
    // Called if the async function throws an error
    console.error('Async function failed:', error)
  }
})

const asyncLimiter = new AsyncRateLimiter(async (id) => {
  await saveToAPI(id)
}, {
  limit: 5,
  window: 1000,
  onExecute: (rateLimiter) => {
    // Called after each successful execution
    console.log('Async function executed', rateLimiter.getExecutionCount())
  },
  onReject: (rateLimiter) => {
    // Called when an execution is rejected
    console.log(`Rate limit exceeded. Try again in ${rateLimiter.getMsUntilNextWindow()}ms`)
  },
  onError: (error) => {
    // Called if the async function throws an error
    console.error('Async function failed:', error)
  }
})

Since the rate limiter's maybeExecute method returns a Promise, you can choose to await each execution before starting the next one. This gives you control over the execution order and ensures each call processes the most up-to-date data. This is particularly useful when dealing with operations that depend on the results of previous calls or when maintaining data consistency is critical.

For example, if you're updating a user's profile and then immediately fetching their updated data, you can await the update operation before starting the fetch.

Just like the synchronous rate limiter, the async rate limiter supports dynamic options for limit, window, and enabled, which can be functions that receive the rate limiter instance. This allows for sophisticated, runtime-adaptive rate limiting behavior.

Each framework adapter provides hooks that build on top of the core async rate limiting functionality to integrate with the framework's state management system. Hooks like createAsyncRateLimiter, useAsyncRateLimitedCallback, or similar are available for each framework.

For core rate limiting concepts and synchronous rate limiting, see the Rate Limiting Guide.