multi contexts and multi pages is not supported

[pakerekchdz]

A single page loads fine, but when I try to load multiple contexts/pages, I get an error. I'm using the latest version, released 8 hours ago, and the standard implementation (as per the docs). Could I get some quick help?

my error:

PS D:\Desktop\browser_test_lightpanda> npx tsx skrypt4.ts
D:\Desktop\browser_test_lightpanda\node_modules\.pnpm\puppeteer-core@24.12.1\node_modules\puppeteer-core\src\cdp\Browser.ts:205
    const {browserContextId} = await this.#connection.send(
           ^


TypeError: Cannot destructure property 'browserContextId' of '(intermediate value)' as it is undefined.
    at CdpBrowser.createBrowserContext (D:\Desktop\browser_test_lightpanda\node_modules\.pnpm\puppeteer-core@24.12.1\node_modules\puppeteer-core\src\cdp\Browser.ts:205:12)
    at async Promise.all (index 1)
    at <anonymous> (D:\Desktop\browser_test_lightpanda\skrypt4.ts:13:18)

Node.js v20.12.0

server error:

$time=1752234963974 $scope=app $level=error $msg="server read error" err=ConnectionResetByPeer

my code:

import puppeteer from 'puppeteer-core';

const browser = await puppeteer.connect({
  browserWSEndpoint: "ws://:9222",
});

const travelSites = [
  "https://www.booking.com",
  "https://www.tripadvisor.com",
  "https://www.expedia.com"
]

const contexts = await Promise.all([
  browser.createBrowserContext(),
  browser.createBrowserContext(),
  browser.createBrowserContext()
]);

// Process multiple pages simultaneously
const results = await Promise.all(
  contexts.map(async (context, index) => {
    const page = await context.newPage();
    await page.goto(travelSites[index]);
    return await page.title();
  })
);

console.log(results);

[pakerekchdz]

@krichprollsch @francisbouvier @nrigaudiere
Is there a version that will simply work? If so, please provide a link. Because, quite simply, nothing works in the new version. Context doesn't work when I create multiple pages, and the page doesn't work when I create multiple pages.

[pakerekchdz]

Please explain how the cloud works for you, but when I run it on Linux, nothing works. It either crashes when adding page/context, or a page crashes with the error "Navigating frame was detached." How on earth does it work for you in the cloud?

This is unusable; I can't use it practically.

[pakerekchdz]

Under the hood it uses chrome, instead of lightpanda xD you have to add "?browser=lightpanda&". After adding that, absolutely nothing works.

[krichprollsch]

Hello @pakerekchdz,

1. Lightpanda doesn't handle multi-contexts/multi-pages. That's why the script you shared doesn't work.
   Please change the script to request pages one by one.

2. The browser is under active development and complex websites (or protected by anti block) like the ones you listed can easily fail or crash.
   It's also possible new issues happens in future versions b/c adding new Web APIs can change the behavior for each website.

I get results with booking but tripadvisor and expedia seems to block.

3. Yes, the cloud version exposes chrome browser by default, you have to set the parameter to use lightpanda.
   It's something we will change in the future.

While we appreciate your feedback, it would be more useful if you provide more specific error and use case.

[pixelkoduje]

It's not about use cases; practically every website gets blocked, so how are we supposed to use this software? It's useless at this point, because sceaping when 99% of websites are blocked is pointless. Even if it would use 50 times less memory.

[pixelkoduje]

Or, to put it another way, how long do you think it will take to upload 100 random pages, add a timer to wait for the page to load and download something without throwing an error? How long do you think such a scenario is likely? Otherwise, it's pointless if it blocks every connection. If I wanted to use it like a circus, I'd use curl or some other lightweight software.

[pixelkoduje]

I mean, how long will it take you to get this working properly, plus or minus? Are we talking a month or six months?