-
-
Save paulrouget/5262478 to your computer and use it in GitHub Desktop.
| <!DOCTYPE html> | |
| <meta charset=utf-8 /> | |
| <title>Defining JavaScript functions, the ES6 way</title> | |
| <h1>Defining JavaScript functions, the ES6 way</h1> | |
| <em>Use Firefox 22 (Firefox Nightly)</em> | |
| <script> | |
| // old school | |
| var square = function(x) { | |
| return x * x; | |
| } | |
| console.log(square(3)); | |
| // Expression closures | |
| // doc: http://mzl.la/10TNpzc | |
| // *Edit:* I've been told that Expression Closures are not part of ES6. Still cool. | |
| var square = function(x) x * x; | |
| console.log(square(3)); | |
| // Arrow functions | |
| // spec: http://bit.ly/KN3z1c | |
| var square = x => { return x * x; }; | |
| console.log(square(3)); | |
| // Arrow function + expression closure | |
| var square = x => x * x; | |
| console.log(square(3)); | |
| // Destructuring assignment | |
| // doc: http://mzl.la/Xed4Ua | |
| var squareAndCube = x => [x*x, x*x*x]; | |
| var [s,c] = squareAndCube(3); | |
| console.log(s + ", " + c); | |
| </script> |
Having multiple options to write the same thing will only decrease readability.
Instead of spending 2 seconds to write the "function()" string, you will spend more time to read code, on the long run.
So, I'm not happy at all about the new ES6 syntax.
Regarding the last example, squareAndCube, it seems like something I might try but later refactor into more traditional code after I return to it confused by what it's doing.
The Mozilla docs cite swapping/rotating multiple variables as a benefit to this type of function. Perhaps. However, when I end up needing to swap values in my own code it typically points to a problem in my design that can be fixed elsewhere.
To me it looks like a "Pitman" shorthand version of JavaScript. It clearly can cut down file sizes, but also makes it more difficult to read and interpret. That said, it looks to be a good way to re-factor code for production, but I'd want to have the same code saved in the traditional format also.
In C# we call example 3 and 4 Lambda Expressions.
Everyone saying that "writing function isn't that much of a problem" clearly is missing the point of "why shorter function syntax." Yes, writing "function ... return" isn't the problem, the problem is when you try to use the paradigm the language was supposed to (according to Brendan) support: JS is a wannabe functional language that doesn't make functional programming easy.
Compare:
var square = function(x){ return x * x }
var add = function(a){ return function(b){ return a + b }}
var compose = function(f){ return function(g) { return function(){ return f(g.apply(this, arguments)) }}}
var squarePlus2 = compose(add(2))(square)
squarePlus2(4) // => 18With:
var square = x => x * x
var add = a => b => a + b
var compose = f => g => (...xs) => f(g.apply(this, xs))
var squarePlus2 = compose(add(2))(square)
squarePlus2(4) // => 18With (Haskell):
square x = x * x
add a b = a + b
compose f g xs = f (g xs)
squarePlus2 = compose (add 2) square
squarePlus2 4 -- => 18
@paulrouget Just a heads up, expression closures are not part of ES6 https://gist.github.com/paulrouget/5262478#file-functions-html-L17-L20
But curly-free, auto returning expression bodies for arrow functions are part of ES6. Basically example 2 (expression closures) isn't a standard part of ES6 but all the other examples are.
Hope I can still use CoffeeScript to handle this..
Arrows are not meant to replace normal functions, u can't use bind,call or apply to change the 'this', the 'this' value will be lexically bound from the function the arrow was declared inside.
I also think arrow functions don't have constructors.
This is very different to how a regular function works, although at first glance they may seem very similar.
@boban984: I was curious about your statements so I went and checked (Chrome v52). Let's see:
let wrapper = {
method1: function() { return this },
method2() { return this },
method3: () => this,
method4: function() { return (() => this)() },
method5: function() { return (() => this) },
method6: () => (() => this)(),
method7: () => (() => this),
}
// all the following are true
wrapper.method1() === wrapper
wrapper.method2() === wrapper
wrapper.method3() === window
wrapper.method4() === wrapper
wrapper.method5()() === wrapper
wrapper.method6() === window
wrapper.method7()() === window
// of particular interest:
let wm4 = wrapper.method4.bind('wow')
let wm5 = wrapper.method5.bind('nice')
`${wm4()} ${wm5()()}` === 'wow nice'I'm not sure why it acts like this, I thought this would always be window in an arrow function.
I also think arrow functions don't have constructors.
You don't need those, because you have classes now. I'm not sure everyone was fond of prototypal inheritance (though personally I liked it).
Moreover, you can always use traditional functional notation when you want the way it behaves. This new fat arrow notation makes Javascript more productive for functional programming (and brings it closer to languages like Haskell, while also having OOP features), and anyone who thinks it's not readable might want to stick to mantaining WordPress blog-sites and using drag-and-drop GUIs with big buttons.
Examples 2 and 4 remind me of Scala. I may be in a minority but I don't find defining old-school functions as much of a pain. It's okay, really!