Why is there a `null` value in JavaScript?

In JavaScript, there are two values which basically say 'I don't exist' - undefined and null.

A property to which a programmer has not assigned anything will be undefined, but in order for a property to become null, null must be explicitly assigned to it.

I once thought that there was a need for null because undefined is a primitive value and null an object. It's not, even if typeof null will yield 'object': Actually, both are primitive values - which means neither undefined nor null can be returned from a constructor function, as both will be converted to an empty object (one has to throw an error to proclaim failure in constructors).

They also both evaluate to false in boolean contexts. The only real difference I can think of is that one evaluates to NaN, the other to 0 in numeric contexts.

So why is there both undefined and null if this just confuses programmers who are incorrectly checking for null when trying to find out whether a property has been set or not?

What I'd like to know is if anyone has a reasonable example where it's necessary to use null which couldn't be expressed using undefined instead.

So the general consensus seems to be that undefined means 'there is no such property' while null means 'the property does exist, but holds no value'.

I could live with that if JavaScript implementations would actually enforce this behavior - but undefined is a perfectly valid primitive value, so it can easily be assigned to existing properties to break this contract. Therefore, if you want to make sure if a property exists, you have to use the in operator or hasOwnProperty() anyway. So once again: what's the practical use for separate values for undefined and null?

I actually use undefined when I want to unset the values of properties no longer in use but which I don't want to delete. Should I use null instead?

The question isn't really "why is there a null value in JS" - there is a null value of some sort in most languages and it is generally considered very useful.

The question is, "why is there an undefined value in JS". Major places where it is used:

1. when you declare var x; but don't assign to it, x holds undefined;
2. when your function gets fewer arguments than it declares;
3. when you access a non-existent object property.

null would certainly have worked just as well for (1) and (2)*. (3) should really throw an exception straight away, and the fact that it doesn't, instead of returning this weird undefined that will fail later, is a big source of debugging difficulty.

*: you could also argue that (2) should throw an exception, but then you'd have to provide a better, more explicit mechanism for default/variable arguments.

However JavaScript didn't originally have exceptions, or any way to ask an object if it had a member under a certain name - the only way was (and sometimes still is) to access the member and see what you get. Given that null already had a purpose and you might well want to set a member to it, a different out-of-band value was required. So we have undefined, it's problematic as you point out, and it's another great JavaScript 'feature' we'll never be able to get rid of.

> I actually use undefined when I want to unset the values of properties no longer in use but which I don't want to delete. Should I use null instead?

Yes. Keep undefined as a special value for signaling when other languages might throw an exception instead.

null is generally better, except on some IE DOM interfaces where setting something to null can give you an error. Often in this case setting to the empty string tends to work.

Best described http://saladwithsteve.com/2008/02/javascript-undefined-vs-null.html">here</a>;, but in summary:

undefined is the lack of a type and value, and null is the lack of a value.

Furthermore, if you're doing simple '==' comparisons, you're right, they come out the same. But try ===, which compares both type and value, and you'll notice the difference.

I don't think there's any reason to have both null and undefined, because the only reason many people have suggested ("undefined means there's no such variable/property") is not valid, at least in JavaScript. undefined can't tell you whether the variable/property exists or not.

console.log(foo);               // "ReferenceError: foo is not defined"
                                // foo does not exist
var foo;
console.log(foo);               // "undefined", a different response
console.log(foo === undefined); // "true", but it does exist

var obj = {};
console.log(obj.hasOwnProperty("foo")); // "false", no such property
obj.foo = undefined;
console.log(obj.hasOwnProperty("foo")); // "true", it exists and has the value "undefined"
console.log(obj.foo === undefined);     // "true", but it does exist

obj.bar = "delete me";
obj.bar = undefined;
console.log(obj.hasOwnProperty("bar")); // "true", not actually deleted
delete obj.bar;
console.log(obj.hasOwnProperty("bar")); // "false", deleted

As you can see, checking foo === undefined does not tell you whether foo exists, and setting obj.bar = undefined does not actually delete bar.

It may be the JavaScript author's original intent that undefined should represent "nonexistence". However, the implementation didn't turn out that way.

It is entirely possible to need both. For instance if you query WMI it is entirely possible to have a class return properties that have a null value. They are defined, they just happen to hold null at the time.

I think that your conclusion that JavaScript defines undefined as "there is no such property" and null as "the property has no value" is perfectly correct. And in a language as dynamic as JavaScript it is a very important distinction. The use of duck typing means that we need to be able to differentiate between a property not existing and not having a value. It is our primary means of deriving type information. in a statically typed language there is a definite distinction between a field being null and a field not existing. In JavaScript this is no different. However it is checked at runtime, and can be modified up until that time.

I'm going to have to agree that the implementation is strange as a lot of time the distinction is blurred. However I think that in JavaScript the distinction is important. And being able to assign undefined is essential.

I remember reading a blog post a while ago about an online RPG written in JavaScript. It used examples where objects were created as copies of existing instances rather than prototypes (classes, functions, whatever), and were then altered. This really made me understand how powerful that undefined could be when modifying existing objects, but I cannot remember who wrote it.

Semantically they mean different things. The type null has exactly one value in its domain, null and a property may be assigned this specific value. Undefined represents a distinct lack of any value having been assigned.

As a Java programmer, I see a huge difference between undefined and null. Coding JavaScript, not so much, because JavaScript is not strongly typed, and the differences between undefined and null are blurred by the automatic conversions that are frequently performed by run-time. BTW, I take frequent advantage of those conversions; they make my JS code more compact and readable.

To answer your question, undefined means that a value was never set. Practically speaking, this generally points to a bug. If yourObject.property is undefined, that means you didn't set the property for some reason, or I'm looking for something that doesn't exist at all. This is a real issue when working on a project with more than one coder.

null means that "no value" was explicitly set. Practically speaking, you are telling me something about the property, perhaps that it isn't used in this context, or that a value has yet to be determined.

In Java, attempts to access a field that is undefined will always result in an exception. In fact, the compiler can be made to warn you about this in your code.

Try this example:

<html>
<head>
    <script type="text/javascript">
        function ShowObjProperties(obj) {
            var property, propCollection = "";

            for(property in obj) {
                propCollection += (property + ": " + obj[property] + "\n");
            }
            
            alert(propCollection);
        }
        
        var obj = {
            userid: 3,
            name: 'me!',
            speak: function() { alert('Hi! My name is ' + this.name + ' and my ID is ' + this.userid + '.'); }
        }
        
        //Shows all properties
        ShowObjProperties(obj);
        
        //The Speak function is no longer in the list!
        delete obj.speak;
        alert(typeof obj.speak);
        ShowObjProperties(obj);
        
        //The UserID is still listed, it just has no value!
        obj.userid = null;
        ShowObjProperties(obj);
    </script>
</head>
<body>
    
</body>
</html>

I think that there is a very real use for 2 different types here.

What it comes down to is javascripts dynamic nature.

Things may be undefined so that they can be added at a later point. This is arguably why javascript is so powerful and extensible.

it's an important language feature if you wrap your program around the evented paradigm of javascript.

// a key exists as a placeholder for something
if(obj[name] === null) 
// no key exists int eh hashtable that is in this object
if(obj[name] === undefined)

this is very useful if you are dealing with a set of data that needs a value to represent 'nothing' to indicate some action different from using 'nothing' to indicate the default action.

filter_func_pt = {
  date:function(d){ return Math.round(d.getTime()/1000);},
  user: null,
}

function transform(obj){
    var ret = {};
    for( var prop in obj){
       var f = filter_func_pt[prop];
       if(f)
          ret[prop] = f(obj);
       else if(filter_func_pt[prop] === null)
         continue;
       else
         ret[prop] == obj;
    }
  return ret;
}

var a = {
   date: new Date(),
   user: 'sam'
   votes: [23, 41, 55] 
};

var b = transform(a);

/* b = {
 *    date: 1298582417
 *    votes: [23, 41, 55]
 * }
 */

in the above code the null keyword and the undefined server very clear and different purposes. the lookup that is not found in the filter_func_pt object which returns undefined means to add the property to the return object as is, whereas a null value indicates that the value should be withheld, and not added, and the presence of any true value in this case represents a function used to transform the value before adding it to the ret object.

null is beautiful

as are all other Types of Live Script.

> cite: In JavaScript, there are two values which basically say 'I don't > exist' - undefined and null.

Why would you want to say incorrect things?!

"null" is "empty Object" just like "0" is an "empty Number". 0, is nothing -yet it exists as a Type of a Number thing. null is of course also empty but "it is" and it's a well defined thing of an Object Type.

It common to speak of these thing as of "types", when they are not. In fact they are "categories". But that's over now.

So will stick to it to say that "null" is a Type of Object without a Kind. And "null" says "I very much exist[!], but I don't have content of my kind".

Whereas undefined lacks both the Type and the Kind where undefined happens to also be its Type definition. An undefined type of a type becomes its distinctive typology. A sort of a [does "nothing" exist and how do you define "nothing"?] question.

> cite: undefined nor null can be returned from a constructor function, > as both will be converter to an empty object

You've managed to once again say an incorrect thing. Of course not, the "undefined" is not an Object, its a plain Token which we humans understand; but contrary to that null is - and it's telling you that: its Type is correct, but the Kind you are looking for is not contained within it, or at least -not at this time. Come visit us later when we put\assign some object in\to it.

> cite: The only real difference I can think of is that one evaluates to > NaN, the other to 0 in numeric contexts.

That makes the whole point of their core distinction, as mentioned: undefined is a plain token and since it is made up of the same 'genetic' material as its distant relatives: strings, the [ +undefined ] operation will duck-convert it to NaN, similarly null will of course morph itself into a correct kind 0 \Number instead, and as opposed to undefined that will morph into a string (!which is not empty!) and that's exactly why it yields NaN instead. Where: +undefined >> +"undefined" >> NaN. Since numeric context expects explicit value.

While the Boolean context expects reference - finds nothing to convert and yields 'false'.

Let's cut through now...

> cite: So once again: what's the practical use for seperate values for > undefined and null?

I will try to give you only two empirical examples and hope to suffice

oElement.onclick >> null

//means -the property exists; its expected value is of Type: Object, and that oElement supports the "onclick" event!

oElement.innerText >> ""

//means - the property exists; its expected value is of Type: String, which means that oElement supports the "innerText" property.

on both cases -if you receive "undefined" it means that the property doesn't exist; is not supported or has a wrong (ua vendor) implementation.

Stay frost and have fun.

after reading amazing discussion regarding undefined vs null, little search on google took me to Mozilla Documentations https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/null it is mentioned - null is often retrieved in a place where an object can be expected but no object is relevant.

Isn't is similar to Null object pattern https://en.wikipedia.org/wiki/Null_object_pattern

So i guess this make sense to have Null data type.

Documentation also mentioned as typeof null // "object" (not "null" for legacy reasons)

Not sure what legacy reasons are