A flexible RxJS operator library that enables polling on any completed observable source with advanced timing and retry strategies.
- Two polling types:
repeatandintervalto suit different use cases - Timing strategies:
constant,linear,exponential,randomanddynamic(custom logic) - Auto-pause: Automatically pause/resume polling based on page visibility (browser only)
- Flexible retries: Control retry attempts with consecutive or total counting modes
- Input validation: Guards against unexpected input time values
- Cross-platform: Works in both browser and Node.js environments
- Modern compatibility: Compatible with RxJS v7+
- Multiple module formats: Supports CJS, ESM, and UMD
npm install rxjs-poll --savePolling is essential when you need to repeatedly check for updates from sources that don't provide real-time notifications. Common scenarios include monitoring HTTP API endpoints for status changes, watching DOM elements for state updates, or periodically sampling data streams.
This operator cleanly separates polling concerns from your core observable logic. It waits for your source observable to complete per polling type, then schedules the next poll based on your configuration. The architecture distinguishes between normal polling delays and error retry scenarios, giving you precise control over both success and failure timing strategies.
Plug and play - just add the operator to your pipe and start polling.
import { poll } from 'rxjs-poll';
import { takeWhile } from 'rxjs';
request$
.pipe(
poll(), // Poll every second with exponential retry strategy (7s max)
takeWhile(({ status }) => status !== 'done', true)
)
.subscribe({ next: console.log });Use built-in strategies for easy timing control.
import { poll } from 'rxjs-poll';
import { takeWhile } from 'rxjs';
request$
.pipe(
poll({
type: 'interval', // Drops uncompleted source after delay
delay: {
strategy: 'random',
time: [1000, 3000], // Random delay between 1 and 3 seconds
},
retry: {
limit: Infinity, // Will never throw
},
}),
takeWhile(({ status }) => status !== 'done', true)
)
.subscribe({ next: console.log });Implement complex polling strategies with dynamic timing based on poll state.
import { poll } from 'rxjs-poll';
import { takeWhile } from 'rxjs';
request$
.pipe(
poll({
delay: {
/** Adaptive polling based on response data */
strategy: 'dynamic',
time: ({ value }) => (value?.items.length ? 1000 : 500),
},
retry: {
/** Custom exponential backoff with jitter */
strategy: 'dynamic',
time: ({ consecutiveRetryCount }) => {
const exponential = Math.pow(2, consecutiveRetryCount - 1) * 1000;
const jitter = Math.random() * 200;
return exponential + jitter;
},
limit: 6,
},
}),
takeWhile(({ status }) => status !== 'done', true)
)
.subscribe({ next: console.log });Creates a polling operator that will begin polling once the source observable completes.
interface PollConfig {
/**
* Defines the polling behavior:
* - 'repeat': Polls after current source completes
* - 'interval': Polls in intervals, dropping any ongoing source operations
* @default 'repeat'
*/
type?: 'repeat' | 'interval';
/**
* Configuration for polling delays (between successful operations)
*/
delay?: {
/**
* Strategy type for delay timing. Built-in strategies (except dynamic)
* calculate time per state's `pollCount`.
* @default 'constant'
*/
strategy: 'constant' | 'random' | 'dynamic';
/**
* Time (ms) depending on strategy:
* - constant: number
* - random: [min, max]
* - dynamic: (state) => number | [min, max]
* @default 1000
*/
time:
| number
| [min: number, max: number]
| (state: PollState) => number | [min: number, max: number];
};
/**
* Configuration for retry behavior (on errors).
* Can be fully specified with strategy/time, or partially specified with just limit/consecutiveOnly.
* If omitted entirely, uses default exponential retry strategy.
* - Full config: `{ strategy: '...', time: ..., limit?: ..., consecutiveOnly?: ... }`
* - Partial config: `{ limit: ..., consecutiveOnly?: ... }` (uses default strategy/time)
* - Omit entirely: Uses all defaults
*/
retry?: {
/**
* Strategy type for retry timing. Built-in strategies (except dynamic)
* calculate time per state:
* - consecutiveOnly: true → uses `consecutiveRetryCount`
* - consecutiveOnly: false → uses `retryCount`
* @default 'exponential'
* @note Required if `time` is provided, otherwise omitted to use defaults
*/
strategy: 'constant' | 'linear' | 'exponential' | 'random' | 'dynamic';
/**
* Time (ms) depending on strategy:
* - constant: number
* - linear: number
* - exponential: number
* - random: [min, max]
* - dynamic: (state) => number | [min, max]
* @default 1000
* @note Required if `strategy` is provided, otherwise omitted to use defaults
*/
time:
| number
| [min: number, max: number]
| (state: PollState) => number | [min: number, max: number];
/**
* Maximum number of retry attempts before throwing an error.
* Use `Infinity` to keep retrying indefinitely.
* @default 3
*/
limit?: number;
/**
* Controls how retries are counted:
* - true: Only consecutive errors count toward retry limit
* (resets counter on success)
* - false: All errors count toward retry limit regardless of
* successful responses between them
* @default true
*/
consecutiveOnly?: boolean;
};
/**
* [Browser only] Controls polling behavior when page isn't visible
* - true: Pause polling when tab isn't active, and resume on active
* - false: Poll even when tab isn't focused
* @default true
*/
pauseWhenHidden?: boolean;
}State object passed to delay/retry time producer functions:
interface PollState<T> {
/** Latest value from the source. For `interval` polling type,
* first emission is undefined. */
value: T | undefined;
/** Latest error when retrying */
error: any | undefined;
/** Total number of successful poll operations */
pollCount: number;
/** Total number of retry attempts */
retryCount: number;
/** Current number of consecutive retry attempts */
consecutiveRetryCount: number;
}
/** Note: pollCount + retryCount = total attempts */Version 2 (source) introduces an API focused on strategy-based configuration with improved type safety and clearer separation between polling delays and retry behavior. See V2 Changes for detailed migration guide and examples.
Version 1 (source) will continue to receive bug fixes and security updates.
Contributions are welcome! Please read our Contributing Guide for details on our code of conduct and the process for submitting pull requests.
Have a question? Check out GitHub Discussions to ask questions, share ideas, or get help from the community.
This library is inspired by rx-polling, which creates an Observable for polling.