[Bug]: chromium channel 'bleeds' into Firefox options

[Ionaru]

Version

1.52.0

Steps to reproduce

1. Set up playwright as normal: npm init playwright@latest
2. Set up 2 projects in playwright.config.js:

projects: [
  {
    name: "chromium",
    use: { ...devices["Desktop Chrome"], channel: "chromium" },
  },
  {
    name: "firefox",
    use: { ...devices["Desktop Firefox"] },
  },
],

3. Write a test for each browser:

import { test, expect, chromium, firefox } from "@playwright/test";

test("Run a test - chromium", async () => {
  const browser = await chromium.launch();
  const context = await browser.newContext();
  const page = await context.newPage();
  expect(await page.evaluate(() => 1 + 1)).toBe(2);
});

test("Run a test - firefox", async () => {
  const browser = await firefox.launch();
  const context = await browser.newContext();
  const page = await context.newPage();
  expect(await page.evaluate(() => 1 + 1)).toBe(2);
});

4. Run the tests using npx playwright test

Expected behavior

Both tests pass.

Actual behavior

The test "Run a test - firefox" fails with this error:

Error: browserType.launch: Unsupported firefox channel "chromium"

Additional context

In the console output, the "Run a test - firefox" test has a [chromium] tag, meaning the chromium project is active, but that should not prevent me from launching a Firefox browser.

Environment

System:
    OS: Windows 11 10.0.26100
    CPU: (32) x64 AMD Ryzen 9 3950X 16-Core Processor
    Memory: 10.75 GB / 31.93 GB
  Binaries:
    Node: 22.14.0 - ~\AppData\Local\nvs\default\node.EXE
    Yarn: 1.22.19 - ~\AppData\Local\nvs\default\yarn.CMD
    npm: 11.3.0 - ~\AppData\Local\nvs\default\npm.CMD
    pnpm: 8.15.1 - ~\AppData\Local\nvs\default\pnpm.CMD
    bun: 1.1.43 - ~\.bun\bin\bun.EXE
  IDEs:
    VSCode: 1.99.3 - C:\Users\NAME\scoop\apps\vscode\current\bin\code.CMD
  Languages:
    Bash: 5.0.17 - C:\Windows\system32\bash.EXE
  npmPackages:
    @playwright/test: ^1.52.0 => 1.52.0
    playwright: ^1.52.0 => 1.52.0

[mxschmitt]

This sounds expected, you should never use launch() inside the test and let Playwright launch the browsers for you. We expose fixtures for browser, context, and page: https://playwright.dev/docs/test-fixtures

If you manually launch a specific browser inside your test it basically means you bypass our existing project infrastructure which is supposed to parameterise tests with different browsers.

Would that work for you?

[Ionaru]

That makes sense, thank you for the answer.