[WPT] use `Undefined` to signal missing param

[sadym-chromium]

Moved from https://github.com/web-platform-tests/wpt/issues/44068

The current WebDriver BiDi WPT tests use None to represent two distinct cases for parameters:

• The parameter should be excluded from the transmission.
• The parameter is not expected in the result.

This overloaded use of None can cause confusion and hinder testing scenarios where we need to pass an explicit null value.

Example:

• Testing bidi_session.browsing_context.set_viewport with missing viewport.

Proposal:

• New Code: In newly developed WebDriver BiDi tests, use the value Undefined to specifically indicate that a parameter should be missing. Use value None to specify the parameter should be present with the value null.
• Existing Code: Phase out the use of None in existing tests (where it is used to represent missing parameters) and replace it with Undefined.

Advantages:

• Clarity: Undefined explicitly expresses the intention of an omitted parameter.
• Testability: This change will allow us to implement specific tests to validate the behavior when a null value is passed as a parameter.

[OrKoN]

I think WPT was the right place for the issue. I am in favor of using Undefined and None as used in browsingContext.setViewport. Alternatively, we could also add a Null type to be used in tests to reflect the CDDL type.

[sadym-chromium]

I think WPT was the right place for the issue.

According to @jgraham the issue better suits this repo.

[whimboo]

Hm, but what should we define in the BiDi spec? Any code change has to be made in wpt anyway. And Undefined in the spec is represented by the optional (?) field. Or what do I miss?

[OrKoN]

I don't think there is an action point for the spec as well.

[sadym-chromium]

The proposal is only about WPT tests

[jgraham]

I think this is the right discussion venue because it's a WG decision about test style, not a test infra decision. Not all issues here have to result in spec changes, but we might, for example, decide to document the agreed best practice in a readme.

[whimboo]

Beside the discussion where the actual discussion should happen I would also vote for a constant style across all of our tests.

[css-meeting-bot]

The Browser Testing and Tools Working Group just discussed WPT use Undefined to signal missing param.

The full IRC log of that discussion

<AutomatedTester> Topic: WPT use Undefined to signal missing param
<AutomatedTester> github: https://github.com/w3c/webdriver-bidi/issues/642
<AutomatedTester> sadym (IRC): This is a technical issue with wpt. Currently we use null for sending things over the wire. This is not always correct.
<orkon> q+
<AutomatedTester> ... sometimes we use the special undefined when we want to test [missed]
<whimboo> q+
<orkon> q-
<whimboo> Python can actually not handle null and undefined with None
<AutomatedTester> ... my suggest is that we move to using a unified way to show undefined when new tests are written when fields. shouldnt be sent
<AutomatedTester> q?
<AutomatedTester> ack next
<AutomatedTester> whimboo: I agree we need somet
<AutomatedTester> ...thing better here. I mentioned this before.
<AutomatedTester> ... my initial proposal is we use null instead of undefined we make something that makes reading the spec and the tests the same
<AutomatedTester> sadym (IRC): I was played around with something that was `optional`
<AutomatedTester> whimboo: would this work as optional equal to none?
<AutomatedTester> ... as none and null are not equal
<AutomatedTester> sadym (IRC): that would work too
<AutomatedTester> ... the only issue is that none is serialised to null in JS
<AutomatedTester> whimboo: let's see if we can play with these in a test and see what works out best
<AutomatedTester> q?