Editorial: Minimize use of the term "callable"

[jmdyck]

In the ECMAScript spec, the term "callable" (as a noun) is synonymous with "function", but that might not be obvious to the reader. (There are languages in which callables are a superset of functions.)

This PR mostly eliminates the word "callable". (I left a few occurrences, e.g. "a function is a callable object" is an informal way to relate concepts.)

The first commit just renames IsCallable to IsFunction. The second commit handles other occurrences. They can probably be squashed for merging.

[ljharb]

I think this is a lot of churn for not a lot of benefit - and this change would be problematic if we ever are able to fix the mess that is [[Call]] vs [[Construct]] in the spec - iow, imo things that you can't invoke without new should be typeof function but shouldn't be considered "callable".

[jmdyck]

this change would be problematic if we ever are able to fix the mess that is [[Call]] vs [[Construct]] in the spec - iow, imo things that you can't invoke without new should be typeof function but shouldn't be considered "callable".

Yeah, if that ever happens then this PR would probably be mostly churn.

Is it likely to happen? I don't see a proposal for it.

[ljharb]

It’s not necessarily normative, and no, there’s nothing pending.

I’m also saying i think it’s churn as-is - proxies aren’t functions but can be callable, as is document.all.

[jmdyck]

proxies aren’t functions but can be callable, as is document.all.

So you're saying there are currently things that are callable but that aren't functions. That's interesting. What is your definition of 'callable'? (or maybe, what's your definition of 'function'?)

It looks to me like callable proxies and document.all are objects that support the [[Call]] internal method, and so are functions, according to the spec definition.

[ljharb]

If i can’t Function.prototype.toString.call it, it’s not a function, regardless of what typeof says.

[bakkot]

You can Function.prototype.toString.call both a Proxy for a function and document.all (and a Proxy for document.all, for that matter).

[ljharb]

ha, fair enough then.

i still don’t think this PR’s churn is worth it, personally.

[michaelficarra]

Discussed at the editor call today. We're going to explore the viability of removing always-throwing [[Call]] and [[Construct]] implementations before deciding which route to take editorially. If it's possible to do that without normative consequences, that would be preferred.

[bathos]

@michaelficarra is there somewhere one could learn more about that / see the discussion?

[bathos]

Re: churn, outside ES repos it looks like IsCallable is used by at least the following specs & proposals:

HTML
Web Audio
Web IDL
WebAssembly JSI
WebAssembly JSI Exception Handling
WebDriver
WebDriver BiDi

also CSS Layout API Level 1, but it looks like it’s not used in the real(?) spec (“lasted editor’s draft” in w3c parlance), just its commit history (“latest published version”).

It’s also a term that’s made its way into userland JS a good amount, I think, but I suppose that wouldn’t have much bearing. A bit sad to see it go myself but admittedly can offer no explanation as to why :)

Horrible thing 👻

You can Function.prototype.toString.call both a Proxy for a function and document.all (and a Proxy for document.all, for that matter).

It even “supports instanceof” with a simple shim.

document.all(0).id = "prototype";
Object.create(document.all(0)) instanceof document.all; // true

[ljharb]

@bathos although https://npmjs.com/is-callable is usually a transitive dependency, the term definitely has some adoption.

[github-actions]

The rendered spec preview for this PR is available as a single page at https://tc39.es/ecma262/pr/2889 and as multiple pages at https://tc39.es/ecma262/pr/2889/multipage .