fetch() plus the policy production code needsTimeouts, retries, typed errors, streaming and security limits, from one Rust core that behaves the same in browsers, Web Workers and Node.js.
Every failure rejects with a typed Error (HttpError with the status and body, TimeoutError, NetworkError, CancelledError…), so one catch handles them all.
timeoutMs covers each attempt up to the last byte of the body. Retries back off with jitter, honour Retry-After, never repeat a POST or PATCH unless you allow it, and can share a client-wide retry budget.
Block private networks (SSRF, checked on the resolved address, for requests and WebSockets), cap response, line and event sizes, drop credentials on cross-origin redirects, refuse HTTPS downgrades.
Shed load beyond a pending-request limit, hold slow servers back with backpressure, and keep a few kilobytes per pending request, measured and budgeted in CI.
ReadableStream bodies, newline-delimited JSON and Server-Sent Events for LLM APIs, upload and download progress, and WebSockets with reconnection on the same client.
The request policy is Rust compiled to WebAssembly. A conformance suite runs the same 52 behaviours on node:http, the native addon, fetch() and real browsers, and CI fails on any undocumented difference.
import { WasmClient } from 'rust-fetch';
const api = new WasmClient({
baseUrl: 'https://api.example.com/v1',
timeoutMs: 10_000, // per attempt, including reading the body
retries: 3, // exponential backoff with jitter; honours Retry-After
});
try {
const { body: user } = await api.get('/users/42');
} catch (error) {
if (error.kind === 'HttpError' && error.status === 404) {
// not found: error.body has the response text
} else {
throw error;
}
}