HTTP servers

import { Nodepod } from '@scelar/nodepod';

let markServerReady: (() => void) | undefined;
const serverReady = new Promise<void>((resolve) => {
  markServerReady = resolve;
});

const nodepod = await Nodepod.boot({
  files: {
    '/server.js': `
      const http = require('http');
      http.createServer((_request, response) => {
        response.end('hello from nodepod');
      }).listen(3000);
    `,
  },
  onServerReady: (port) => {
    if (port === 3000) markServerReady?.();
  },
});

await nodepod.spawn('node', ['server.js']);
await serverReady;

spawn() waits for the process worker to be ready, not for an HTTP server inside that process to finish listening. Use onServerReady (as above) before calling request() or port().

Call the server without a service worker:

const response = await nodepod.request(3000, { path: '/' });
console.log(new TextDecoder().decode(response.body));

For navigable browser previews, use nodepod.port(3000) after configuring the service worker and preview bridge. Production hosts should provide a dedicated wildcard origin:

const nodepod = await Nodepod.boot({
  previewOrigin: 'https://{instanceId}-{port}.preview.example.com',
});

Dedicated preview origins separate cookies and service-worker scope from the host application. They do not make untrusted code harmless; review the security model.