65

I'm using "Closure Compiler", when compiling my scripts I spend the following:

Before compiling:

// ==ClosureCompiler==
// @compilation_level SIMPLE_OPTIMIZATIONS
// @output_file_name default.js
// @formatting pretty_print,print_input_delimiter
// ==/ClosureCompiler==

var myObj1 = (function() {

  var undefined;   //<----- declare undefined

  this.test = function(value, arg1) {

    var exp = 0;
    arg1 = arg1 == undefined ? true : arg1;  //<----- use declare undefined
    exp = (arg1) ? value * 5 :  value * 10;

    return exp;
  };

  return this;
}).call({});

var myObj2 = (function() {

  this.test = function(value, arg1) {

    var exp = 0;
    arg1 = arg1 == undefined ? true : arg1;  //<----- without declare undefined
    exp = (arg1) ? value * 5 :  value * 10;

    return exp;
  };

  return this;
}).call({});

Compiled:

// Input 0
var myObj1 = function() {
  this.test = function(b, a) {
    a = a == void 0 ? true : a;  //<-----
    var c = 0;
    return c = a ? b * 5 : b * 10
  };
  return this
}.call({}), myObj2 = function() {
  this.test = function(b, a) {
    a = a == undefined ? true : a; //<-----
    var c = 0;
    return c = a ? b * 5 : b * 10
  };
  return this
}.call({});

With this I believe the question of the use of "void 0 " and "undefined", is there any difference in the use or the two cases are well?.

Edit

if I define "var undefined" compiled with "void 0 ", if I did not define "undefined" compiled with "undedined. " then not a matter of number of characters between "undefined" and "void 0"

Test

Edit II: performance, based on this link

Code and Test

IE 8:
typeof: 228ms
undefined: 62ms
void 0: 57ms

Firefox 3.6:
typeof: 10ms
undefined: 3ms
void 0: 3ms

Opera 11:
typeof: 67ms
undefined: 19ms
void 0: 20ms

Chrome 8:
typeof: 3ms
undefined: 5ms
void 0: 3ms

andres descalzo
  • 14,887
  • 13
  • 64
  • 115

4 Answers4

74

From MDN:

The void operator evaluates the given expression and then returns undefined.

This operator allows inserting expressions that produce side effects into places where an expression that evaluates to undefined is desired.

The void operator is often used merely to obtain the undefined primitive value, usually using "void(0)" (which is equivalent to "void 0"). In these cases, the global variable undefined can be used instead (assuming it has not been assigned to a non-default value).

Closure Compiler swaps in void 0 because it contains fewer characters than undefined, therefore producing equivalent, smaller code.


Re: OP comment

yes, I read the documentation, but in the example I gave, "google closure" in a case using "void 0" and another "undefined"

I believe this is actually a bug in Google Closure Compiler!

Community
  • 1
  • 1
Matt Ball
  • 354,903
  • 100
  • 647
  • 710
56

The real only semantic difference between void expr and undefined is that on ECMAScript 3, the undefined property of the global object (window.undefined on browser environments) is writable, whereas the void operator will return the undefined value always.

A popular pattern that is often implemented, to use undefined without worries is simply declaring an argument, and not passing anything to it:

(function (undefined) {
  //...
  if (foo !== undefined) {
    // ...
  }

})();

That will allow minifiers to shrink the argument maybe to a single letter (even shorter than void 0 :), e.g.:

(function (a) {
  //...
  if (foo !== a) {
    // ...
  }
})();
Christian C. Salvadó
  • 807,428
  • 183
  • 922
  • 838
  • Nice. Thanks for that, I was wondering whether the `undefined` value is a property or not. http://typeofnan.blogspot.com/2011/01/typeof-is-fast.html – jAndy Jan 26 '11 at 15:44
  • @jAndy, Nice article, and thanks for the mention ;), BTW you could add a test against `void 0`, it might be interesting... – Christian C. Salvadó Jan 26 '11 at 15:53
  • Thanks! I insta checked that, `void 0` performs way better than `undefined` obviously now, but it's still behind the cached version. – jAndy Jan 26 '11 at 15:59
  • I think if you use this method on any non-anonymous and not immediately called functions you'll be asking for trouble. If somebody calls it with the parameter (in most cases passing extra parameters to a function doesn't influence behavior, such as `alert('test', '2nd param ignored')`) for example through code like `someFunc.apply(this, arguments)` you risk comparing against a completely different variable. I wouldn't rely on this technique to save a mere 5 characters that'll most likely be even less when GZIPed. – Aidiakapi Apr 15 '14 at 11:03
  • CMS, but `window.undefined = 0` but `assert( 0 != undefined)`. Am I missing something. – Alexander Suraphel Apr 20 '15 at 07:04
  • Stay away from this! You can't see if you're passing a variable to undefined or not because the actual call can be 100's of lines away. While `foo !== undefined` will probably still work when you're passing in an other value for `undefined`, tests like `foo === undefined` will open pandora's box. – huysentruitw Sep 17 '15 at 17:38
10

Just a follow-up on all the answers before.

They look the same, but to the Compiler they are completely different.

The two code sections compile to different outputs because one is referring to a local variable (the var undefined), and the compiler simply in-lines it because it is used exactly once and is no more than one line. If it is used more than once, then this in-lining won't happen. The in-lining provides a result of "undefined", which is shorter to represent as "void 0".

The one without a local variable is referring to the variable called "undefined" under the global object, which is automatically "extern'ed" by the Closure Compiler (in fact, all global object properties are). Therefore, no renaming takes place, and no in-lining takes place. Voila! still "undefined".

Stephen Chung
  • 14,497
  • 1
  • 35
  • 48
5

There is no difference, Try it yourself:

void 0 === undefined

will evaluate to true.
undefined is 3 characters longer, I guess that is the reason why they use it that way.

jAndy
  • 231,737
  • 57
  • 305
  • 359
  • 2
    And therefore it's a bandwidth optimization: send fewer bytes over the wire? – Joel Coehoorn Jan 26 '11 at 15:30
  • 1
    @JoelCoehoorn: well the closure compiler is a minifier aswell, I guess they try to squeeze every byte here. – jAndy Jan 26 '11 at 15:31
  • if I define "var undefined" compiled with "void 0 ", if I did not define "undefined" compiled with "undedined". then not a matter of number of characters between "undefined" and "void 0" – andres descalzo Jan 26 '11 at 15:38
  • 2
    `(function () { return 'foo' })() === 'foo'` also returns true. DOES THAT MEAN THERE IS NO DIFFERENCE BETWEEN THE TWO? – Harry Dec 20 '16 at 07:47
  • 2
    @Harry, in the code there is a huge difference. But after evaluation, both sides are actually "foo" strings. Undefined is the result of void. So, even if void 0, in the code, is not the same as undefined, it really evaluates to undefined, it is undefined after evaluation. In the same way we say "2+2 is 4". Yes, "2+2" is NOT 4, it evaluates to 4, but after evaluation it IS 4! – FrancescoMM Sep 15 '17 at 16:29
  • There's no difference *if undefined has not been redefined*. That's not always a safe assumption. – Steve Bennett Jan 10 '20 at 03:12