Tanstack Pacer

Overview

TanStack Pacer provides a unified, type-safe toolkit for controlling function execution timing. It offers class-based APIs, factory functions, and React hooks for debouncing, throttling, rate limiting, queuing, and batching.

Core: @tanstack/pacer React: @tanstack/react-pacer Status: Beta

Installation

npm install @tanstack/pacer
npm install @tanstack/react-pacer  # React hooks

Debouncing

Delays execution until after a period of inactivity.

Class API

import { Debouncer } from '@tanstack/pacer'

const debouncer = new Debouncer(
  (query: string) => fetchSearchResults(query),
  {
    wait: 300,            // ms of inactivity before execution
    leading: false,       // Execute on leading edge (default: false)
    trailing: true,       // Execute on trailing edge (default: true)
    maxWait: 1000,        // Force execution after 1s of continuous calls
    enabled: true,
    onExecute: (result) => console.log(result),
  }
)

debouncer.maybeExecute('search term')
debouncer.cancel()
debouncer.getExecutionCount()
debouncer.setOptions({ wait: 500 }) // Dynamic reconfiguration

Factory Function

import { debounce } from '@tanstack/pacer'

const debouncedSearch = debounce(
  (query: string) => fetchResults(query),
  { wait: 300 }
)

debouncedSearch('term')
debouncedSearch.cancel()

React Hooks

import {
  useDebouncer,
  useDebouncedCallback,
  useDebouncedState,
  useDebouncedValue,
} from '@tanstack/react-pacer'

// Full debouncer instance
const debouncer = useDebouncer(fn, { wait: 300 })

// Simple debounced function
const debouncedFn = useDebouncedCallback(fn, { wait: 300 })

// Debounced state management
const [debouncedValue, setValue] = useDebouncedState(initialValue, { wait: 300 })

// Debounced reactive value
const debouncedValue = useDebouncedValue(reactiveValue, { wait: 300 })

Throttling

Limits execution to at most once per interval.

Class API

import { Throttler } from '@tanstack/pacer'

const throttler = new Throttler(
  (position: { x: number; y: number }) => updatePosition(position),
  {
    wait: 100,            // Minimum interval between executions
    leading: true,        // Execute immediately on first call (default: true)
    trailing: true,       // Execute after interval with last args (default: true)
    enabled: true,
    onExecute: (result) => console.log(result),
  }
)

throttler.maybeExecute({ x: 100, y: 200 })
throttler.cancel()

React Hooks

import {
  useThrottler,
  useThrottledCallback,
  useThrottledState,
  useThrottledValue,
} from '@tanstack/react-pacer'

const throttledFn = useThrottledCallback(handleScroll, { wait: 100 })
const [throttledPos, setPos] = useThrottledState({ x: 0, y: 0 }, { wait: 100 })

Rate Limiting

Controls execution with a maximum count within a time window.

Class API

import { RateLimiter } from '@tanstack/pacer'

const limiter = new RateLimiter(
  async (endpoint: string) => fetch(endpoint).then(r => r.json()),
  {
    limit: 10,            // Max executions per window
    window: 60000,        // Time window in ms (60s)
    enabled: true,
    onExecute: (result) => console.log(result),
    onReject: (...args) => console.warn('Rate limited:', args),
  }
)

limiter.maybeExecute('/api/data')  // Rejected if limit exceeded
limiter.getExecutionCount()
limiter.getRejectionCount()

React Hooks

import {
  useRateLimiter,
  useRateLimitedCallback,
  useRateLimitedState,
  useRateLimitedValue,
} from '@tanstack/react-pacer'

const rateLimitedFn = useRateLimitedCallback(apiCall, { limit: 5, window: 1000 })

Queuing

Sequential execution with configurable concurrency.

import { Queue } from '@tanstack/pacer'

const queue = new Queue({
  concurrency: 1,         // Max concurrent tasks
  started: true,          // Start processing immediately
})

queue.add(() => uploadFile(file1))
queue.add(() => uploadFile(file2))

queue.start()
queue.pause()
queue.clear()
queue.getSize()           // Pending count
queue.getPending()        // Currently executing count

Batching

Groups calls for combined processing.

import { Batcher } from '@tanstack/pacer'

const batcher = new Batcher(
  (items: LogEntry[]) => sendBatchToServer(items),
  {
    maxSize: 50,          // Auto-flush at 50 items
    wait: 1000,           // Auto-flush after 1s
  }
)

batcher.add(logEntry)    // Accumulates
batcher.flush()          // Manual flush
batcher.getSize()        // Current batch size
batcher.clear()          // Discard batch

Async Variants

import { AsyncDebouncer, asyncDebounce, AsyncThrottler, asyncThrottle } from '@tanstack/pacer'

const asyncDebouncer = new AsyncDebouncer(
  async (query: string) => {
    const response = await fetch(`/api/search?q=${query}`)
    return response.json()
  },
  { wait: 300 }
)

// React async hooks
import { useAsyncDebouncer, useAsyncThrottler } from '@tanstack/react-pacer'

Choosing the Right Utility

| Scenario | Utility | Why | |----------|---------|-----| | Search input | Debouncer | Wait for user to stop typing | | Scroll events | Throttler | Periodic updates during activity | | API protection | RateLimiter | Hard limit on call frequency | | File uploads | Queue | Sequential processing | | Analytics events | Batcher | Group for efficiency | | Network requests | AsyncDebouncer | Handle abort/retry |

Leading vs Trailing Edge

• Leading (leading: true): Execute immediately, suppress until wait expires. Good for button clicks.
• Trailing (trailing: true): Execute after activity stops. Good for search inputs.
• Both: Execute immediately AND after final wait. Good for scroll throttling.

Best Practices

1. Use maxWait with debouncing to guarantee execution during continuous activity
2. Use async variants for network requests (handle abort/cancellation)
3. React hooks handle cleanup automatically - no manual teardown needed
4. Use setOptions for dynamic reconfiguration (e.g., reducing wait for power users)
5. Compose utilities for complex scenarios (rate-limited queue)
6. Use onReject on RateLimiter to inform users when they're rate limited

Common Pitfalls

• Using debounce when you need throttle (debounce waits for inactivity, throttle guarantees periodic execution)
• Not using maxWait with debounce for long-running continuous events
• Creating new instances on every render (use hooks or module-level)
• Forgetting cleanup in non-React environments (call cancel())

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: tanstack-skills
• Source: tanstack-skills/tanstack-skills
• License: MIT

Install and usage instructions live in the source repository linked above.