Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

How do I write a named arrow function in ES2015?

People also ask

Can arrow functions be named?

Like traditional function expressions, arrow functions are not hoisted, and so you cannot call them before you declare them. They are also always anonymous—there is no way to name an arrow function.

What is the use of => in JavaScript?

This is an arrow function. Arrow functions are a short syntax, introduced by ECMAscript 6, that can be used similarly to the way you would use function expressions. In other words, you can often use them in place of expressions like function (foo) {...} . But they have some important differences.


How do I write a named arrow function in ES2015?

You do it the way you ruled out in your question: You put it on the right-hand side of an assignment or property initializer where the variable or property name can reasonably be used as a name by the JavaScript engine. There's no other way to do it, but doing that is correct and fully covered by the specification.

Per spec, this function has a true name, sayHello:

var sayHello = name => {
    console.log(name + ' says hello');
};

This is currently defined in Assignment Operators > Runtime Semantics: Evaluation where it does the abstract NamedEvalution operation (currently step 1.c.i). (You can see everywhere this applies by hovering your mouse over NamedEvalution in the header there and clicking "References".) (Previously, before ES2019, Assignment Operators > Runtime Semantics: Evaluation used the abstract SetFunctionName operation, step 1.e.iii, but in ES2019 onward this specification abstraction was replaced with NamedEvalution.)

Similiarly, PropertyDefinitionEvaluation uses NamedEvalution and thus gives this function a true name:

let o = {
    sayHello: name => {
        console.log(`${name} says hello`);
    }
};

Modern engines set the internal name of the function for statements like that already.

For example, in Chrome, Edge (Chromium-based, v79 onward), or Firefox, open the web console and then run this snippet:

"use strict";
let foo = () => { throw new Error(); };
console.log("foo.name is: " + foo.name);
try {
    foo();
} catch (e) {
    console.log(e.stack);
}

On Chrome 51 and above and Firefox 53 and above (and "Legacy" Edge 13 and above with an experimental flag, or "Chromium" Edge 79 onward), when you run that, you'll see:

foo.name is: foo
Error
    at foo (http://stacksnippets.net/js:14:23)
    at http://stacksnippets.net/js:17:3

Note the foo.name is: foo and Error...at foo.

On Chrome 50 and earlier, Firefox 52 and earlier, and Legacy Edge without the experimental flag, you'll see this instead because they don't have the Function#name property (yet):

foo.name is: 
Error
    at foo (http://stacksnippets.net/js:14:23)
    at http://stacksnippets.net/js:17:3

Note that the name is missing from foo.name is:, but it is shown in the stack trace. It's just that actually implementing the name property on the function was lower priority than some other ES2015 features; Chrome and Firefox have it now; Edge has it behind a flag, presumably it won't be behind the flag a lot longer.

Obviously, I can only use this function after I have defined it

Correct. There is no function declaration syntax for arrow functions, only function expression syntax, and there's no arrow equivalent to the name in an old-style named function expression (var f = function foo() { };). So there's no equivalent to:

console.log(function fact(n) {
    if (n < 0) {
        throw new Error("Not defined for negative numbers");
    }
    return n == 0 ? 1 : n * fact(n - 1);
}(5)); // 120

You have to break it into two expressions (I'd argue you should do that anyway):

const fact = n => {
    if (n < 0) {
        throw new Error("Not defined for negative numbers.");
    }
    return n == 0 ? 1 : n * fact(n - 1);
};
console.log(fact(5));

Of course, if you have to put this where a single expression is required, you can always...use an arrow function:

console.log((() => {
    const fact = n => {
        if (n < 0) {
            throw new Error("Not defined for negative numbers.");
        }
        return n == 0 ? 1 : n * fact(n - 1);
    };
    return fact(5);
})()); // 120

I ain't sayin' that's pretty, but it works if you absolutely, positively need a single expression wrapper.


Side note: What if you don't want a function to get its name from the identifier you're assigning to? That, suppose you don't want example.name to be "example" here?

const example = () => {};
console.log(example.name); // "example"

You can avoid it by using any expression that doesn't use NamedEvaluation. Probably the most popular way to do this sort of thing is the comma operator:

const example = (0, () => {});
//              ^^^−−−−−−−−−^
console.log(example.name); // ""

The 0 there can be anything you want, it's evaluated and then thrown away so 0 is a popular choice. Passing the function through the comma operator breaks the direct link between the assignment and the function expression, preventing NamedEvaluation from providing the name example for the function. (This is similar to other famous uses of the comma operator, like (0, object.example)() which calls object.example without making object the value of this within the call, or (0, eval)("code"), which does an eval, but not in the current scope as it normally would.)

(Thank you to Sebastian Simon for raising this point in the comments.)


No. The arrow syntax is a shortform for anonymous functions. Anonymous functions are, well, anonymous.

Named functions are defined with the function keyword.


If by 'named', you mean you want the .name property of your arrow function to be set, you're in luck.

If an arrow function is defined on the right-hand-side of an assignment expression, the engine will take the name on the left-hand-side and use it to set the arrow function's .name, e.g.

var sayHello = (name) => {
    console.log(name + ' says hello');
}

sayHello.name //=== 'sayHello'

Having said that, your question seems to be more 'can I get an arrow function to hoist?'. The answer to that one is a big ol' "no", I'm afraid.