Consider specifying WebDriver BiDi modules for automated testing

[OrKoN]

With crrev.com/c/6021359 landing for chrome-headless-shell, there is now a way to test web applications with various screen configurations without having the physical devices or resorting to other ways to emulate them on the operating system level. Although this works for the headless shell, there is currently no standard way for developers to test Window Management in major browsers.

Given that it might be difficult to set up a test environment for web apps relying on various screen configurations, I think it would be great if this specification defined how a test interface should look like and how it should integrate with the rest of the implementation.

WebDriver BiDi standard already has some window management capabilities:

• https://w3c.github.io/webdriver-bidi/#command-browser-setClientWindowState
• https://w3c.github.io/webdriver-bidi/#command-browser-getClientWindows

There are a few examples of external specifications extending WebDriver BiDi already:

• https://webbluetoothcg.github.io/web-bluetooth/#automated-testing
• https://www.w3.org/TR/permissions/#automation-webdriver-bidi

[michaelwasserman]

Thanks for pointing that out!

Our hope was to add WebDriver support for configuring OS-level virtual screens via Chromium's VirtualDisplayUtil, where supported.

Otherwise, mocking display configurations for the ScreenDetails API's source-of-truth (or even that of the browser itself) could be of limited use. Past WPTs mocked the API's data, but that pattern unfortunately didn't align with a Chromium re-implementation and the WPTs and mock interface were removed.

See Chromium's multi-screen testing docs: https://crsrc.org/c/docs/ui/display/multiscreen_testing.md
Unfortunately, I'm not aware of an active roadmap to add WPTs or testing guidance and support here.