This allows headers.forEach.length === 1, which matches browsers.

hmm a couple of things here:

1. I can add that for extra clarity but the end result feels the same as thisArg will be undefined when headers.forEach.length === 1 anyways.
2. To match browsers, the default value would be a global object such as window. On node, I guess global would do the job. Try this: [1].forEach(function () { console.log(this === global) })

I can add that for extra clarity

The result indeed is the same inside forEach, but the = undefined is essential to get the correct .length property. Try

console.log(({ f(a, b) {} }).f.length);
console.log(({ f(a, b = undefined) {} }).f.length);

You will see that the first statement prints 2, while the second prints 1. We want the second behavior for forEach, and that's why you should add it. But indeed, the expected behavior is that if thisArg is not supplied, then it defaults to undefined.

By the way, the .length property of the forEach method is defined in the Web IDL standard, see "define the iteration methods" step 2.4.4.

To match browsers, the default value would be a global object such as window.

That's not quite right. In browsers, the default value is window only if we are not in strict mode:

Our test environments are all strict mode, but you can run Reflect.apply(function() { console.log(this); }, undefined, []) in the Node.js REPL (which is sloppy by default) to see that this is actually global in Node.js too.

All this is to say, you don't have to do anything special to get the global object behavior. (In fact, you cannot do anything here to replicate that in pure JavaScript, since there's no easy way of detecting whether a function is strict or sloppy.)

If you're interested in finding where this behavior is specified, it's in the ECMAScript standard, within OrdinaryCallBindThis abstract operation's steps 5–6.

hey @TimothyGu ! Thanks for that! It was very informative. All suggestions are committed now. Let me know if there's anything else I should do.

Btw, there's just a lint issue going on that I'm not able to fix. :/

Can you also test the this value where thisArg is not supplied?

sure! waiting for considerations on #1074 (comment).