JavaScript closures & callback functions - javascript

Can someone clarify this point for me?
Two of my favourite features in JavaScript are closures and the fact that functions are first class objects.
If I want to take advantage of closure in, say, a callback function, then the function must always be an inner function. For example:
doNested();
function doNested() {
var message='Hello';
window.setTimeout(inner, 3000);
function inner() {
alert(message);
}
}
There is no way I can pass a parameter to an outer callback function. For example:
doSeparated();
function doSeparated() {
var message='Goodbye';
window.setTimeout(outer,3000);
}
function outer() {
alert(message);
}
This doesn’t work, of course.
The question is, is there any way to pass inner variables to an external function like this, short of adding them to a parameter list?
Thanks

If I want to take advantage of closure in, say, a callback function, then the function must always be an inner function.
That's correct. Every function in JavaScript only has access those variables which are either defined in its own scope or defined in a parent scope1. Therefore, your first example works while your second example doesn't.
The question is, is there any way to pass inner variables to an external function like this, short of adding them to a parameter list?
No, there's no way to do that. Well, technically you could add your inner variable to an object and then bind that object to the external function after which you can access the inner variable from the this context of the external function, but that is no better than passing the variable to the function directly.
doSeparated();
function doSeparated() {
var message = "Goodbye";
setTimeout(outer.bind({
message: message
}), 3000);
}
function outer() {
alert(this.message);
}
Since you are using setTimeout, you can pass extra arguments to setTimeout which will be given to the callback function. This gets rid of the nasty bind:
doSeparated();
function doSeparated() {
var message = "Goodbye";
setTimeout(outer, 3000, message);
}
function outer(message) {
alert(message);
}
Note that both bind and extra arguments of setTimeout don't work in older versions of IE. In that case, you can use currying instead (which in my humble opinion is the best solution aside from your original nested solution):
doSeparated();
function doSeparated() {
var message = "Goodbye";
setTimeout(outer(message), 3000);
}
function outer(message) {
return function () {
alert(message);
};
}
Other than these, there's no other good solution that I can think of. The best solution is your original nested solution.
1 A function doesn't have access to any variable defined in a child scope or else you would be able to access every variable from the global scope.

Related

About function in javascript

Sorry for the rather beginner question. What's the differences of function usage between
$(document).keypress(function() {
//some code
});
and
var somethingElse = function() {
//some code
};
I know the latter is to create a function similar to Java's
public void somethingElse() {
//some code
}
but I always assume the former as anonymous function that act inside a method. Can someone shed me some light regarding this? Thanks.
The first one is a jQuery shortcut to create an event listener.
It's equivalent to:
document.addEventListener('keypress', function() {
// some code
});
More info: http://www.w3schools.com/jsref/met_document_addeventlistener.asp
Now, about named or anonymous functions, what's the difference between
var doSomething = function() {
// some code
};
and this?
function doSomething() {
// some code
}
There's no difference for the developer. Of course there's a difference on memory, but in javascript developers don't have to take care of it.
Actually for the case of an event handler or other techniques that use callback functions, you can pass an anonymous function or a previously declared one, it's exactly the same:
$(document).keypress(doSomething);
or
$(document).keypress(function() {
// some code
});
creates an anon function and passes it to the handler
creates an anonymous function and a variable that references it.
creates a named function - that is hoisted
the hoisted function becomes available to you at any line within your function scope, while the non-hoisted function will be undefined until the line where it is declared runs.
there is no difference between #2 and #3 (other than the hoisting) - some people think that the first one creates an object and the 2nd one is some magical thing, or a global function, but nope - they are both function objects within your scope.
The former is a callback, meaning some instructions that will be executed ONLY as soon as the keypress event in your example is triggered.
Thus, a function's name is not required.
Function expressions, the latter, is mostly used when adding an object's property acting as a method like:
var myObject = {};
myObject.sayHello = function() {
alert("Hello");
}

JavaScript: invocation as a function/method

I am trying to sharpen my JavaScript skills and I am aware that there are four basic ways to invoke a function - which alter the way this is defined. The two I am interested in are the basic two:
Invocation as a function
Invocation as a method
Which is fine. The first, this will refer to the window object:
function doSomething() {
console.log(this);
}
doSomething(); // window is logged
And the second this will refer to the object it is being executed from within:
var t = {
doSomething: function () {
console.log(this);
}
};
t.doSomething(); // t is logged
Which is all fine. However, is it correct to say, that in these two invoking methods, this is always going to return the object the method is contained within (if that makes sense)?
In the first example, doSomething() is, in reality, defined within the window object - so is a property of the window object, even if we do not define it (or reference it).
Therefore, can it not be said that, in reality, invocation as a function is invocation as a method? Or not?
Imagine defining your function, not in the scope of the global object like you did, but in the scope of another function:
function outer() {
var inner = function(){console.log(this)};
inner();
};
outer();
What do you think the this will be bound to ? What will be printed ?
You might believe it will be the outer function object but its turns out that it will always be the global object.
It is just as if you had written:
function outer() {
var inner = function(){console.log(this)};
inner.call(window);
};
outer();
This is a quite hazardous and far from intuitive JS "feature".
So, to get back to your first example, what you are doing is basically this:
doSomething = function () {
console.log(this);
}
doSomething().call(window);
And, like Yoshi pointed out, if you were using the ES5's strict mode, you would end up doing something like that instead, which is, in my opinion, far less dangerous:
doSomething().call(undefined);
It also gives you a strong hint about the fact that you should never use this in a function ;)

Declare javascript function with extra properties in a single statement?

I know I can define properties on functions, which can then be accessed from within the function. Right now, the only syntax I can work out involves two statements. Is there a more concise way to express the following:
function myFunc() {
// do some work
}
myFunc.myProp = 0;
I'm not looking for a solution that is fewer characters -- this isn't code golf. I'm asking something more along the lines of "are there different patterns of function declaration that have this other desirable merit?" This is almost about ways to use closures (because I suspect the answer lies there).
Thanks!
Especially if you want to access properties of the function from inside the function itself, you're better off doing this:
var theFunction = function() {
function theRealFunction() {
// the code
if (theRealFunction.something == "not whatever")
// do something
// more code
}
theRealFunction.something = "whatever";
return theRealFunction;
}();
What that does is wrap your function declaration up in an anonymous function. The problem with accessing function properties via the function name is that it must do that by finding the function's name in the surrounding scope. That's kind-of icky, but at least this way it involves a scope that's essentially private. It'll work whether or not the resulting function (returned as the return value of the anonymous function) is assigned to a different variable, passed to a function as a handler function, etc.
This really, really all depends. If you're looking for private variables, then you can easily return a function from the function -- an inner-function will contain access to its parent's scope.
var outer_function = (function () {
var private_var = "secret",
public_var = "public",
inner_function = function () {
return private_var;
};
inner_function.public_var = public_var;
return inner_function;
}());
outer_function now equals inner_function, with the benefit of having access to the enclosed data. Any properties attached to the inner (in the way you did) will now be accessible as public properties of outer.
To this end, you can return, say, a constructor for a class, with public-static properties, with the enclosed vars acting as private-static properties, shared between every instance of the "class" you build.
Not exactly the answer to your question, but if you ever want to read up some different design patterns that can be used when defining a javascript function, this is one of the best articles I've ever read on the topic:
http://www.klauskomenda.com/code/javascript-programming-patterns/

Javascript closure

The following program returns "local" and, according to the tutorial Im reading, it is designed to demonstrate the phenomenon ofclosure`
What I don`t understand is why, at the end, in order to call parentfunction, it assigns it to the variable "child" and then calls "child."
Why doesn`t it work by just writing parentFunction(); at the end?
var variable = "top-level";
function parentFunction() {
var variable = "local";
function childFunction() {
print(variable);
}
return childFunction;
}
var child = parentFunction();
child();
parentFunction() returns another function which you assign to var child. Then, you call child() to invoke the function returned by the call to parentFunction().
Running just parentFunction(); at the end wouldn't do anything useful because you would just discard its return value which is a function. But this would work:
parentFunction()();
See this fiddle: http://jsfiddle.net/USCjn/
Update: A simpler example:
function outer() { // outer function returns a function
return function() {
alert('inner function called');
}
}
x = outer(); // what is now in x? the inner function
// this is the same as saying:
// x = function() {
// alert('inner function called');
// }
x(); // now the inner function is called
See this fiddle: http://jsfiddle.net/bBqPY/
Functions in JavaScript can return functions (that can return functions (that can return functions ...)). If you have a function that returns another function then it means that when you call the outer function what you get is the inner function but it is not called yet. You have to call the value that you got as a function to actually run the body of the inner function. So:
x = f();
means - run a function f and store what it returns (which may be a string, a number, an object, an array, or a function) in x. But this:
x = f()();
means - run a function f, expect it to return a function and run that returned function as well (the second parentheses) and store in x what the returned function returned.
The function f here is a higher order function because it returns another function. Functions can also take another functions as arguments. One of the most powerful ideas of functional programming languages in general and JavaScript in particular is that functions are just normal values like arrays or numbers that can be returned and passed around.
You have to first grasp the idea of higher order functions to understand closures and the event system in JavaScript.
2016 Update
Note that currently this:
function outer() {
return function() {
alert('inner function called');
}
}
can be written as:
let outer = () => () => alert('inner function called');
using the ES6 arrow function syntax.
The amazing part about closures is that an inner function (in this case, childFunction) can refer to variables outside of its scope (in this case, variable). parentFunction doesn't return the result of childFunction, but an actual reference to the function!
This means that when you do the following...
var child = parentFunction();
...now, child has a reference to childFunction, and childFunction still has access to any variables it had when the function was created, even if they no longer exist.
In order to have parentFunction call childFunction, you'd need to change your code as follows:
From...
return childFunction;
To:
return childFunction();
Douglas Crockford (Pioneer of JSON, among other things) has a whole article devoted to closures, and scoping in javascript, and it would be well worth it to check out his other articles on javascript.
The point that is being demonstrated is that the function that was returned and assigned to child is still referencing the variable that was declared inside parentFunction instead of the one that was declared outside where child() is being invoked.
The only way to create a variable scope in javascript is in a function body. Normally the variable inside the parentFunction would have been discarded after the function returned.
But because you declared a function inside parentFunction that referenced variable in that scope and passed it out of parentFunction, the variable in the parentFunction is retained via the reference made in the new function.
This protects variable from outside manipulation except by functions that closed around it inside parentFunction.

What's the meaning of "()" in a function call?

Now, I usually call a function (that requires no arguments) with () like this:
myFunction(); //there's empty parens
Except in jQuery calls where I can get away with:
$('#foo').bind('click', myFunction); //no parens
Fine. But recently I saw this comment here on SO:
"Consider using setTimeout(monitor, 100); instead of setTimeout('monitor()', 100);. Eval is evil :)"
Yikes! Are we really eval()-ing a string here? I guess I don't really understand the significance and implications of 'calling' a function. What are the real rules about calling and referring to functions?
In JavaScript functions are first-class objects. That means you can pass functions around as parameters to a function, or treat them as variables in general.
Let's say we are talking about a function hello,
function hello() {
alert('yo');
}
When we simply write
hello
we are referring to the function which doesn't execute it's contents. But when we add the parens () after the function name,
hello()
then we are actually calling the function which will alert "yo" on the screen.
The bind method in jQuery accepts the type of event (string) and a function as its arguments. In your example, you are passing the type - "click" and the actual function as an argument.
Have you seen Inception? Consider this contrived example which might make things clearer. Since functions are first-class objects in JavaScript, we can pass and return a function from within a function. So let's create a function that returns a function when invoked, and the returned function also returns another function when invoked.
function reality() {
return function() {
return function() {
alert('in a Limbo');
}
};
}
Here reality is a function, reality() is a function, and reality()() is a function as well. However reality()()() is not a function, but simply undefined as we are not returning a function (we aren't returning anything) from the innermost function.
So for the reality function example, you could have passed any of the following to jQuery's bind.
$('#foo').bind('click', reality);
$('#foo').bind('click', reality());
$('#foo').bind('click', reality()());
Your jQuery bind example is similar to setTimeout(monitor, 100);, you are passing a reference of a function object as an argument.
Passing a string to the setTimeout/setInterval methods should be avoided for the same reasons you should avoid eval and the Function constructor when it is unnecessary.
The code passed as a string will be evaluated and run in the global execution context, which can give you "scope issues", consider the following example:
// a global function
var f = function () {
alert('global');
};
(function () {
// a local function
var f = function() {
alert('local');
};
setTimeout('f()', 100); // will alert "global"
setTimeout(f, 100); // will alert "local"
})();
The first setTimeout call in the above example, will execute the global f function, because the evaluated code has no access to the local lexical scope of the anonymous function.
If you pass the reference of a function object to the setTimeout method -like in the second setTimeout call- the exact same function you refer in the current scope will be executed.
You are not doing the same thing in your jQuery example as in the second setTimeout example - in your code you are passing the function and binding the click event.
In the first setTimout example, the monitor function is passed in and can be invoked directly, in the second, the sting monitor() is passed in and needs to be evaled.
When passing a function around, you use the function name. When invoking it, you need to use the ().
Eval will invoke what is passed in, so a () is required for a successful function invocation.
First of all, "()" is not part of the function name.
It is syntax used to make function calls.
First, you bind a function to an identifier name by either using a function declaration:
function x() {
return "blah";
}
... or by using a function expression:
var x = function() {
return "blah";
};
Now, whenever you want to run this function, you use the parens:
x();
The setTimeout function accepts both and identifier to a function, or a string as the first argument...
setTimeout(x, 1000);
setTimeout("x()", 1000);
If you supply an identifier, then it will get called as a function.
If you supply an string, than it will be evaluated (executed).
The first method (supplying an identifier) is preferred ...

Categories