The code there is quite outdated, and should be using webidl converters. There are many examples of such across the codebase, including the same file:

From the spec, it states that the signal property can be null.

signal (null or an [AbortSignal] object)

From the spec, it states that the signal property can be null.

signal (null or an [AbortSignal] object)

@randomicon00 I think null is not the javascript null value, it's more like undefined on spec text.

there is a wpt test which states a TypeError should throw when signal is null.

https://github.com/web-platform-tests/wpt/blob/3532a4b4082fbbb32438ae8670252e664aaa2efe/dom/events/AddEventListenerOptions-signal.any.js#L135-L138

From the spec, it states that the signal property can be null.

signal (null or an [AbortSignal] object)

If anything, the code snippet that I provide yields different results than Deno when evaluated in a browser (do try it yourself).
I expect that the null is internal, and intended to mark handlers that aren't storing an AbortSignal reference, but it isn't intended to accept null in the exposed interface as an AbortSignal object.

@Garcia101 what's your deno version? "AddEventListenerOptions.signal shouldn't be nullable" fix is landed at version v1.12.0， so if you using version lower than version 1.12.0, then the problem will exist.

fix pr: #11348
release note: https://github.com/denoland/deno/releases/tag/v1.12.0

I am using version 1.20.6, the problem is not exist

$ cat test.js                                          
const et = new EventTarget();
et.addEventListener("foo", () => {}, { signal: null });

$ deno run test.js
error: Uncaught TypeError: signal must be non-null
et.addEventListener("foo", () => {}, { signal: null });
   ^
    at EventTarget.addEventListener (deno:ext/web/02_event.js:939:15)
    at file:///Users/feng/dev/deno/test.js:2:4

Deno Version

$ deno --version   
deno 1.20.6 (release, x86_64-apple-darwin)
v8 10.0.139.6
typescript 4.6.2

@Garcia101 what's your deno version? "AddEventListenerOptions.signal shouldn't be nullable" fix is landed at version v1.12.0， so if you using version lower than version 1.12.0, then the problem will exist.

fix pr: #11348 release note: https://github.com/denoland/deno/releases/tag/v1.12.0

I am using version 1.20.6, the problem is not exist

$ cat test.js                                          
const et = new EventTarget();
et.addEventListener("foo", () => {}, { signal: null });

$ deno run test.js
error: Uncaught TypeError: signal must be non-null
et.addEventListener("foo", () => {}, { signal: null });
   ^
    at EventTarget.addEventListener (deno:ext/web/02_event.js:939:15)
    at file:///Users/feng/dev/deno/test.js:2:4

Deno Version

$ deno --version   
deno 1.20.6 (release, x86_64-apple-darwin)
v8 10.0.139.6
typescript 4.6.2

I just ran this, to be fully certain:

const event_target = new EventTarget();

event_target.addEventListener("abc", null, {
 signal: null
});

console.log(Deno.version);

output:

{ deno: "1.20.6+0e4574b", v8: "10.0.139.6", typescript: "4.6.2" }

I also tried on the non-canary release,

{ deno: "1.20.6", v8: "10.0.139.6", typescript: "4.6.2" }

Furthermore, note that I reference what happens to be the code intended to prevent this bug, in the issue message itself:

From a quick skim, it seems to be that this branch isn't being reached, even though it's intended to be?

deno/ext/web/02_event.js

Lines 932 to 934 in 8ae1702

	} else if (options?.signal === null) {
	throw new TypeError("signal must be non-null");
	}

That very code coincidentally comes from the PR that was intended to fix this issue.

The current implementation is erroneous and misaligned with the appropriate specification, as WebIDL type conversions are up-front, whereas this code is evaluated conditionally, that is, if the handler callback function is not null. Please do evaluate my suggested script snippet in your current version of Deno, and you will see the same. CrowlCats agreed that this is the case. So hopefully it may be fixed in due time. (Unfortunately, I don't have a means of locally building and testing Deno at the time, so my apologies for not attempting this myself.)

@Garcia101 my bad. I copy the snippet from the wpt test. I can reproduce this problem when I execute your snippet.