Can someone clarify the difference between a constructor function and a factory function in Javascript.
When to use one instead of the other?
The basic difference is that a constructor function is used with the new keyword (which causes JavaScript to automatically create a new object, set this within the function to that object, and return the object):
var objFromConstructor = new ConstructorFunction();
A factory function is called like a "regular" function:
var objFromFactory = factoryFunction();
But for it to be considered a "factory" it would need to return a new instance of some object: you wouldn't call it a "factory" function if it just returned a boolean or something. This does not happen automatically like with new, but it does allow more flexibility for some cases.
In a really simple example the functions referenced above might look something like this:
function ConstructorFunction() {
this.someProp1 = "1";
this.someProp2 = "2";
}
ConstructorFunction.prototype.someMethod = function() { /* whatever */ };
function factoryFunction() {
var obj = {
someProp1 : "1",
someProp2 : "2",
someMethod: function() { /* whatever */ }
};
// other code to manipulate obj in some way here
return obj;
}
Of course you can make factory functions much more complicated than that simple example.
One advantage to factory functions is when the object to be returned could be of several different types depending on some parameter.
Benefits of using constructors
Most books teach you to use constructors and new
this refers to the new object
Some people like the way var myFoo = new Foo(); reads.
Drawbacks
Details of instantiation get leaked into the calling API (via the new requirement), so all callers are tightly coupled to the constructor implementation. If you ever need the additional flexibility of the factory, you'll have to refactor all callers (admittedly the exceptional case, rather than the rule).
Forgetting new is such a common bug, you should strongly consider adding a boilerplate check to ensure that the constructor is called correctly ( if (!(this instanceof Foo)) { return new Foo() } ). EDIT: Since ES6 (ES2015) you can't forget new with a class constructor, or the constructor will throw an error.
If you do the instanceof check, it leaves ambiguity as to whether or not new is required. In my opinion, it shouldn't be. You've effectively short circuited the new requirement, which means you could erase drawback #1. But then you've just got a factory function in all but name, with additional boilerplate, a capital letter, and less flexible this context.
Constructors break the Open / Closed Principle
But my main concern is that it violates the open/closed principle. You start out exporting a constructor, users start using the constructor, then down the road you realize you need the flexibility of a factory, instead (for instance, to switch the implementation to use object pools, or to instantiate across execution contexts, or to have more inheritance flexibility using prototypal OO).
You're stuck, though. You can't make the change without breaking all the code that calls your constructor with new. You can't switch to using object pools for performance gains, for instance.
Also, using constructors gives you a deceptive instanceof that doesn't work across execution contexts, and doesn't work if your constructor prototype gets swapped out. It will also fail if you start out returning this from your constructor, and then switch to exporting an arbitrary object, which you'd have to do to enable factory-like behavior in your constructor.
Benefits of using factories
Less code - no boilerplate required.
You can return any arbitrary object, and use any arbitrary prototype - giving you more flexibility to create various types of objects which implement the same API. For example, a media player that can create instances of both HTML5 and flash players, or an event library which can emit DOM events or web socket events. Factories can also instantiate objects across execution contexts, take advantage of object pools, and allow for more flexible prototypal inheritance models.
You'd never have a need to convert from a factory to a constructor, so refactoring will never be an issue.
No ambiguity about using new. Don't. (It will make this behave badly, see next point).
this behaves as it normally would - so you can use it to access the parent object (for example, inside player.create(), this refers to player, just like any other method invocation would. call and apply also reassign this, as expected. If you store prototypes on the parent object, that can be a great way to dynamically swap out functionality, and enable very flexible polymorphism for your object instantiation.
No ambiguity about whether or not to capitalize. Don't. Lint tools will complain, and then you'll be tempted to try to use new, and then you'll undo the benefit described above.
Some people like the way var myFoo = foo(); or var myFoo = foo.create(); reads.
Drawbacks
new doesn't behave as expected (see above). Solution: don't use it.
this doesn't refer to the new object (instead, if the constructor is invoked with dot notation or square bracket notation, e.g. foo.bar() - this refers to foo - just like every other JavaScript method -- see benefits).
A constructor returns an instance of the class you call it on. A factory function can return anything. You would use a factory function when you need to return arbitrary values or when a class has a large setup process.
A Constructor function example
function User(name) {
this.name = name;
this.isAdmin = false;
}
let user = new User("Jack");
new creates an object prototyped on User.prototype and calls User with the created object as its this value.
new treats an argument expression for its operand as optional:
let user = new User;
would cause new to call User with no arguments.
new returns the object it created, unless the constructor returns an object value, which is returned instead. This is an edge case which for the most part can be ignored.
Pros and Cons
Objects created by constructor functions inherit properties from the constructor's prototype property, and return true using the instanceOf operator on the constructor function.
The above behaviors can fail if you dynamically change the value of the constructor's prototype property after having already used the constructor. Doing so is rare, and it can't be changed if the constructor were created using the class keyword.
Constructor functions can be extended using the extends keyword.
Constructor functions can't return null as an error value. Since it's not an object data type, it is ignored by new.
A Factory function example
function User(name, age) {
return {
name,
age,
}
};
let user = User("Tom", 23);
Here the factory function is called without new. The function is entirely responsible for the direct or indirect use if its arguments and the type of object it returns. In this example it returns a simple [Object object] with some properties set from arguments.
Pros and Cons
Easily hides the implementation complexities of object creation from the caller. This is particularly useful for native code functions in a browser.
The factory function need not always return objects of the same type, and could even return null as an error indicator.
In simple cases, factory functions can be simple in structure and meaning.
Objects returned do not generally inherit from the factory function's prototype property, and return false from instanceOf factoryFunction.
The factory function can't be safely extended using the extends keyword because extended objects would inherit from the factory functions prototype property instead of from the prototype property of the constructor used by the factory function.
Factories are "always" better. When using object orientated languages then
decide on the contract (the methods and what they will do)
Create interfaces that expose those methods (in javascript you don't have interfaces so you need to come up with some way of checking the implementation)
Create a factory that returns an implementation of each interface required.
The implementations (the actual objects created with new) are not exposed to the factory user/consumer. This means that the factory developer can expand and create new implementations as long as he/she doesn't break the contract...and it allows for the factory consumer to just benefit from the new API without having to change their code...if they used new and a "new" implementation comes along then they have to go and change every line which uses "new" to use the "new" implementation...with the factory their code doesn't change...
Factories - better than all anything else - the spring framework is completely built around this idea.
Factories are a layer of abstraction, and like all abstractions they have a.cost in complexity. When encountering a factory based API figuring out what the factory is for a given API can be challenging for the API consumer. With constructors discoverability is trivial.
When deciding between ctors and factories you need to decide if the complexity is justified by the benefit.
Worth noting that Javascript constructors can be arbitrary factories by returning something other than this or undefined. So in js you can get the best of both worlds - discoverable API and object pooling/caching.
I think the factory function is superior to the constructor function. Using new with the constructor function, we are binding our code to one specific way of creating an object, while with a factory, we are free so we can create more different instances without binding ourselves. Let's say we have this class:
const file = new CreateFile(name)
If we want to refactor CreateFile class, creating subclasses for the file format our server supports, we can write an elegan factory function:
function CreateFile(name) {
if (name.match(/\.pdf$/)) {
return new FilePdf(name);
} else if (name.match(/\.txt$/)) {
return new FileTxt(name);
} else if (name.match(/\.md$/)) {
return new FileMd(name);
} else {
throw new Error("Not supprted file type");
}
}
with factory functions, we can implement private variables, hide the information from the users which is called encapsulation.
function createPerson(name) {
const privateInfo = {};
// we create person object
const person = {
setName(name) {
if (!name) {
throw new Error("A person must have a name");
}
privateInfo.name = name;
},
getName() {
return privateInfo.name;
},
};
person.setName(name);
return person;
}
For the differences, Eric Elliott clarified very well,
But for the second question:
When to use one instead of the other?
If you are coming from the object-oriented background, Constructor function looks more natural to you.
this way you shouldn't forget to use new keyword.
I have a function set up to return a wrapped C++ object when called as
new MyClass();
but I want to also be able to say
MyClass.do_something();
I know how to do what I want in pure javascript:
MyClass.prototype = { do_something: function(){}};
but how do I do the same in C++?
I'm aware of the InstanceTemplate() and PrototypeTemplate() methods on v8::FunctionTemplate, but those seem to only be used in the creation of the new object returned when new MyClass() is called. How do I get at the actual function's prototype?
Thank you.
I saw this post, but am not sure if it's relevant: Add a function template to a global object prototype in v8
Turns out I was way overthinking the problem.
You simply call .Set() on your v8::FunctionTemplate and pass in another v8::FunctionTemplate as the value.
my_constructor_function_template.Set("static_method_name", static_method_function_template);
I often use the pattern of a main JavaScript constructor function and adding methods to its prototype object so they can be called intuitively by the user, for example:
function Slideshow(options) {
this.options = options
this.slideshow = $('#slideshow')
//more variables here
}
Slideshow.method1 = function () {
this.slideshow.addClass('test') // do something with slideshow variable
};
Slideshow.method2 = function () {
// another method
};
The one thing that really bugs me about this pattern is how in order to make variables accessible across all prototype methods, I have to add "this" in front of each variable inside the constructor function. It's a major pain, and I can't help but think there's a more elegant way to do this.
If I forgo using the prototype object and just add the methods as instance methods, I know I can't get around this problem, but I like the efficiency? and self encapsulating nature of this pattern. Any other suggestions for a better pattern? Thanks!
It's a major pain
No, it's really not. Every single JavaScript developer uses this syntax. If you were in Ruby or Python, you'd use self., in PHP you'd use $this->. Some languages like C++ don't require any special decorator, but JavaScript does.
and I can't help but think there's a more elegant way to do this.
No, there isn't.
This is JavaScript's syntax, you cannot change it, and you cannot work around it. If you want to access a property of this, you need this. before the property name. Otherwise, you're talking about global variables.
If you want a different syntax, consider a different language like CoffeeScript, which compiles to JavaScript.
meager has pretty much summed things up if you're talking about accessing public instance properties or methods. You have to use this in front of it as that's just how the language works.
But, if you have private instance properties or methods, you can define those as local variables inside the constructor and you can access them without this.
function slideshow(options) {
// no need to resave the options arguments as they can be used
// directly from the argument variable
// define a per-instance private variable that other methods defined
// within the constructor can use directly without the use of `this`
var theShow = $(options.selector || '#slideshow');
// define public methods
this.method1 = function() {
// can access private instance variable here without this in front of it
theShow.addClass('test');
}
this.method2 = function() {
theShow.addClass(options.decoaration);
}
}
This general design pattern is described here: http://javascript.crockford.com/private.html
Practically speaking, this works because the constructor function with the public methods declared inside it creates a closure that lasts for the duration of the object lifetime so the local variables in the constructor become per-instance variables accessible only from the functions declared within the constructor.
So I define classes by using their constructor functions, and in these functions I define all their properties, and do constructor-y things like setting up listeners and such. After defining the constructor, I then go and add all of it's functions directly to its prototype.
To do inheritance, I essentially do this:
namespace.Class = function(param) {
namespace.BaseClass.call(this, param);
//Other constructor stuff
}
namespace.Class.prototype = new namespace.BaseClass();
namespace.Class.constructor = namespace.Class;
The problem is that the BaseClass' constructor gets called when creating the prototype, which causes the prototype to be an instance of that type. This may not be a problem, but it just feels "messy" and performance lagging, and potentially buggy because I have to check if the params are defined.
So my question is: Is it possible to extend a base class without calling its constructor? Or will I have to use tricks such as checking if param is undefined, or using a loop to do stuff?
Sorry if I've missed the whole point of prototypal inheritance, I am being taught Java in uni, and Java doesn't approve of this type of stuff.
Yes it is possible. The preferred way is to use Object.create [MDN]:
namespace.Class.prototype = Object.create(namespace.BaseClass.prototype);
This creates a new, empty object which inherits from BaseClass.prototype, which is all you need at this point.
This is a much cleaner solution than creating an instance of the parent class, because, as you already mentioned, if the super class expects arguments, you don't know which ones to pass.
Which method below is best to define a constructor prototype and why?
Method 1:
MyConstructor.prototype.myFunction1 = function(){};
MyConstructor.prototype.myFunction2 = function(){};
Method 2:
MyConstructor.prototype = {
myFunction1: function(){},
myFunction2: function(){}
};
I'm mostly concerned about speed. Thanks!
I would say there wouldn't be much of a difference. Using an object literal to assign to the Object.prototype is something you can't do if you're assigning the prototype within the constructor (which can be usefull sometimes).
Maybe you should write a little performance test using jsperf.com.
var example = new MyConstructor();
under method 1:
example.constructor === MyConstructor;
under method 2:
typeof(example.constructor) === 'undefined';
The prototype object that comes with a function has a property constructor that points back to the function. If you assign to the proprties of that object, you keep the constructor property. If you overwrite the prototype property with a new object, you lose the constructor property.
The performance difference is minimal. Because constructor is so fragile, you can't really trust it, so I don't bother to preserve it.
You should use the method 1. Using the method 2, everytime you create a new instance, you will "re-create" the methods, since they are inside the constructor.
Speaking further about the readability of your code,
method 1 is better than method 2.
Method 2 spend one more indentation. So it cause difficulty for reading codes.
Additionally, in my case,
I can't inference that whether this function is prototype method or just static member function when we see a function name part in the lower part of codes.
Personally, in conclusion,
I prefer method 2 if there not be much of a difference about performance.
Thanks!