Revision 2.9

Aaron Whyte
Bob Jervis
Dan Pupius
Eric Arvidsson
Fritz Schneider
Robby Walker
This style guide contains many details that are initially hidden from view. They are marked by the triangle icon, which you see here on your left. Click it now. You should see "Hooray" appear below.

Hooray! Now you know you can expand points to get more details. Alternatively, there's a "toggle all" at the top of this document.

JavaScript is the main client-side scripting language used by many of Google's open-source projects. This style guide is a list of dos and don'ts for JavaScript programs.

Declarations with var: Always When you fail to specify var, the variable gets placed in the global context, potentially clobbering existing values. Also, if there's no declaration, it's hard to tell in what scope a variable lives (e.g., it could be in the Document or Window just as easily as in the local scope). So always declare with var. >Use NAMES_LIKE_THIS for constants. Use @const where appropriate. Never use the const keyword.

For simple primitive value constants, the naming convention is enough.

/** * The number of seconds in a minute. * @type {number} */ goog.example.SECONDS_IN_A_MINUTE = 60;

For non-primitives, use the @const annotation.

/** * The number of seconds in each of the given units. * @type {Object.<number>} * @const */ goog.example.SECONDS_TABLE = { minute: 60, hour: 60 * 60 day: 60 * 60 * 24 }

This allows the compiler to enforce constant-ness.

As for the const keyword, Internet Explorer doesn't parse it, so don't use it.

Always use semicolons.

Relying on implicit insertion can cause subtle, hard to debug problems. Don't do it. You're better than that.

There are a couple places where missing semicolons are particularly dangerous:

// 1. MyClass.prototype.myMethod = function() { return 42; } // No semicolon here. (function() { // Some initialization code wrapped in a function to create a scope for locals. })(); var x = { 'i': 1, 'j': 2 } // No semicolon here. // 2. Trying to do one thing on Internet Explorer and another on Firefox. // I know you'd never write code like this, but throw me a bone. [normalVersion, ffVersion][isIE](); var THINGS_TO_EAT = [apples, oysters, sprayOnCheese] // No semicolon here. // 3. conditional execution a la bash -1 == resultOfOperation() || die();
  1. JavaScript error - first the function returning 42 is called with the second function as a parameter, then the number 42 is "called" resulting in an error.
  2. You will most likely get a 'no such property in undefined' error at runtime as it tries to call x[ffVersion][isIE]().
  3. die is called unless resultOfOperation() is NaN and THINGS_TO_EAT gets assigned the result of die().

JavaScript requires statements to end with a semicolon, except when it thinks it can safely infer their existence. In each of these examples, a function declaration or object or array literal is used inside a statement. The closing brackets are not enough to signal the end of the statement. Javascript never ends a statement if the next token is an infix or bracket operator.

This has really surprised people, so make sure your assignments end with semicolons.

Yes

Nested functions can be very useful, for example in the creation of continuations and for the task of hiding helper functions. Feel free to use them.

No

Do not do this:

if (x) { function foo() {} }

While most script engines support Function Declarations within blocks it is not part of ECMAScript (see ECMA-262, clause 13 and 14). Worse implementations are inconsistent with each other and with future EcmaScript proposals. ECMAScript only allows for Function Declarations in the root statement list of a script or function. Instead use a variable initialized with a Function Expression to define a function within a block:

if (x) { var foo = function() {} }
Yes

You basically can't avoid exceptions if you're doing something non-trivial (using an application development framework, etc.). Go for it.

Yes

Without custom exceptions, returning error information from a function that also returns a value can be tricky, not to mention inelegant. Bad solutions include passing in a reference type to hold error information or always returning Objects with a potential error member. These basically amount to a primitive exception handling hack. Feel free to use custom exceptions when appropriate.

Always preferred over non-standards features

For maximum portability and compatibility, always prefer standards features over non-standards features (e.g., string.charAt(3) over string[3] and element access with DOM functions instead of using an application-specific shorthand).

No

There's no reason to use wrapper objects for primitive types, plus they're dangerous:

var x = new Boolean(false); if (x) { alert('hi'); // Shows 'hi'. }

Don't do it!

However type casting is fine.

var x = Boolean(0); if (x) { alert('hi'); // This will never be alerted. } typeof Boolean(0) == 'boolean'; typeof new Boolean(0) == 'object';

This is very useful for casting things to number, string and boolean.

Not preferred

Multi-level prototype hierarchies are how JavaScript implements inheritance. You have a multi-level hierarchy if you have a user-defined class D with another user-defined class B as its prototype. These hierarchies are much harder to get right than they first appear!

For that reason, it is best to use goog.inherits() from the Closure Library or something similar.

function D() { goog.base(this) } goog.inherits(D, B); D.prototype.method = function() { ... };
Foo.prototype.bar = function() { ... };

While there are several methods for attaching methods and properties to a constructor, the preferred style is:

Foo.prototype.bar = function() { /* ... */ };
Yes, but be careful.

The ability to create closures is perhaps the most useful and often overlooked feature of JS. Here is a good description of how closures work .

One thing to keep in mind, however, is that a closure keeps a pointer to its enclosing scope. As a result, attaching a closure to a DOM element can create a circular reference and thus, a memory leak. For example, in the following code:

function foo(element, a, b) { element.onclick = function() { /* uses a and b */ }; }

the function closure keeps a reference to element, a, and b even if it never uses element. Since element also keeps a reference to the closure, we have a cycle that won't be cleaned up by garbage collection. In these situations, the code can be structured as follows:

function foo(element, a, b) { element.onclick = bar(a, b); } function bar(a, b) { return function() { /* uses a and b */ } }
Only for deserialization (e.g. evaluating RPC responses)

eval() makes for confusing semantics and is dangerous to use if the string being eval()'d contains user input. There's usually a better, more clear, safer way to write your code, so its used is generally not permitted. However eval makes deserialization considerably easier than the non-eval alternatives, so its use is acceptable for this task (for example, to evaluate RPC responses).

Deserialization is the process of transforming a series of bytes into an in-memory data structure. For example, you might write objects out to a file as:

users = [ { name: 'Eric', id: 37824, email: 'jellyvore@myway.com' }, { name: 'xtof', id: 31337, email: 'b4d455h4x0r@google.com' }, ... ];

Reading these data back into memory is as simple as evaling the string representation of the file.

Similarly, eval() can simplify decoding RPC return values. For example, you might use an XMLHttpRequest to make an RPC, and in its response the server can return JavaScript:

var userOnline = false; var user = 'nusrat'; var xmlhttp = new XMLHttpRequest(); xmlhttp.open('GET', 'http://chat.google.com/isUserOnline?user=' + user, false); xmlhttp.send(''); // Server returns: // userOnline = true; if (xmlhttp.status == 200) { eval(xmlhttp.responseText); } // userOnline is now true.
No

Using with clouds the semantics of your program. Because the object of the with can have properties that collide with local variables, it can drastically change the meaning of your program. For example, what does this do?

with (foo) { var x = 3; return x; }

Answer: anything. The local variable x could be clobbered by a property of foo and perhaps it even has a setter, in which case assigning 3 could cause lots of other code to execute. Don't use with.

Only in object constructors, methods, and in setting up closures

The semantics of this can be tricky. At times it refers to the global object (in most places), the scope of the caller (in eval), a node in the DOM tree (when attached using an event handler HTML attribute), a newly created object (in a constructor), or some other object (if function was call()ed or apply()ed).

Because this is so easy to get wrong, limit its use to those places where it is required:

  • in constructors
  • in methods of objects (including in the creation of closures)
Only for iterating over keys in an object/map/hash

for-in loops are often incorrectly used to loop over the elements in an Array. This is however very error prone because it does not loop from 0 to length - 1 but over all the present keys in the object and its prototype chain. Here are a few cases where it fails:

function printArray(arr) { for (var key in arr) { print(arr[key]); } } printArray([0,1,2,3]); // This works. var a = new Array(10); printArray(a); // This is wrong. a = document.getElementsByTagName('*'); printArray(a); // This is wrong. a = [0,1,2,3]; a.buhu = 'wine'; printArray(a); // This is wrong again. a = new Array; a[3] = 3; printArray(a); // This is wrong again.

Always use normal for loops when using arrays.

function printArray(arr) { var l = arr.length; for (var i = 0; i < l; i++) { print(arr[i]); } }
Never use Array as a map/hash/associative array

Associative Arrays are not allowed... or more precisely you are not allowed to use non number indexes for arrays. If you need a map/hash use Object instead of Array in these cases because the features that you want are actually features of Object and not of Array. Array just happens to extend Object (like any other object in JS and therefore you might as well have used Date, RegExp or String).

No

Do not do this:

var myString = 'A rather long string of English text, an error message \ actually that just keeps going and going -- an error \ message to make the Energizer bunny blush (right through \ those Schwarzenegger shades)! Where was I? Oh yes, \ you\'ve got an error and all the extraneous whitespace is \ just gravy. Have a nice day.';

The whitespace at the beginning of each line can't be safely stripped at compile time; whitespace after the slash will result in tricky errors; and while most script engines support this, it is not part of ECMAScript.

Yes

Use Array and Object literals instead of Array and Object constructors.

Array constructors are error-prone due to their arguments.

// Length is 3. var a1 = new Array(x1, x2, x3); // Length is 2. var a2 = new Array(x1, x2); // If x1 is a number and it is a natural number the length will be x1. // If x1 is a number but not a natural number this will throw an exception. // Otherwise the array will have one element with x1 as its value. var a3 = new Array(x1); // Length is 0. var a4 = new Array();

Because of this, if someone changes the code to pass 1 argument instead of 2 arguments, the array might not have the expected length.

To avoid these kinds of weird cases, always use the more readable array literal.

var a = [x1, x2, x3]; var a2 = [x1, x2]; var a3 = [x1]; var a4 = [];

Object constructors don't have the same problems, but for readability and consistency object literals should be used.

var o = new Object(); var o2 = new Object(); o2.a = 0; o2.b = 1; o2.c = 2; o2['strange key'] = 3;

Should be written as:

var o = {}; var o2 = { a: 0, b: 1, c: 2, 'strange key': 3 };
No

Modifying builtins like Object.prototype and Array.prototype are strictly forbidden. Modifying other builtins like Function.prototype is less dangerous but still leads to hard to debug issues in production and should be avoided.

No

Don't do this:

var f = function () { /*@cc_on if (@_jscript) { return 2* @*/ 3; /*@ } @*/ };

Conditional Comments hinder automated tools as they can vary the JavaScript syntax tree at runtime.

In general, use functionNamesLikeThis, variableNamesLikeThis, ClassNamesLikeThis, EnumNamesLikeThis, methodNamesLikeThis, and SYMBOLIC_CONSTANTS_LIKE_THIS.

Expand for more information.

  • Private properties, variables, and methods (in files or classes) should be named with a trailing underscore.
  • Protected properties, variables, and methods should be named without a trailing underscore (like public ones).

For more information on private and protected, read the section on visibility .

Optional function arguments start with opt_.

Functions that take a variable number of arguments should have the last argument named var_args. You may not refer to var_args in the code; use the arguments array.

Optional and variable arguments can also be specified in @param annotations. Although either convention is acceptable to the compiler, using both together is preferred.

Getters and setters for properties are not required. However, if they are used, then getters must be named getFoo() and setters must be named setFoo(value). (For boolean getters, isFoo() is also acceptable, and often sounds more natural.)

JavaScript has no inherent packaging or namespacing support.

Global name conflicts are difficult to debug, and can cause intractable problems when two projects try to integrate. In order to make it possible to share common JavaScript code, we've adopted conventions to prevent collisions.

ALWAYS prefix identifiers in the global scope with a unique pseudo namespace related to the project or library. If you are working on "Project Sloth", a reasonable pseudo namespace would be sloth.*.

var sloth = {}; sloth.sleep = function() { ... };

Many JavaScript libraries, including the Closure Library and Dojo toolkit give you high-level functions for declaring your namespaces. Be consistent about how you declare your namespaces.

goog.provide('sloth'); sloth.sleep = function() { ... };

When choosing a child-namespace, make sure that the owners of the parent namespace know what you are doing. If you start a project that creates hats for sloths, make sure that the Sloth team knows that you're using sloth.hats.

"External code" is code that comes from outside your codebase, and is compiled independently. Internal and external names should be kept strictly separate. If you're using an external library that makes things available in foo.hats.*, your internal code should not define all its symbols in foo.hats.*, because it will break if the other team defines new symbols.

foo.require('foo.hats'); /** * WRONG -- Do NOT do this. * @constructor * @extend {foo.hats.RoundHat} */ foo.hats.BowlerHat = function() { };

If you need to define new APIs on an external namespace, then you should explicitly export the public API functions, and only those functions. Your internal code should call the internal APIs by their internal names, for consistency and so that the compiler can optimize them better.

foo.provide('googleyhats.BowlerHat'); foo.require('foo.hats'); /** * @constructor * @extend {foo.hats.RoundHat} */ googleyhats.BowlerHat = function() { ... }; goog.exportSymbol('foo.hats.BowlerHat', googleyhats.BowlerHat);

Use local aliases for fully-qualified types if doing so improves readability. The name of a local alias should match the last part of the type.

/** * @constructor */ some.long.namespace.MyClass = function() { }; /** * @param {some.long.namespace.MyClass} a */ some.long.namespace.MyClass.staticHelper = function(a) { ... }; myapp.main = function() { var MyClass = some.long.namespace.MyClass; var staticHelper = some.long.namespace.MyClass.staticHelper; staticHelper(new MyClass()); };

Do not alias namespaces.

myapp.main = function() { var namespace = some.long.namespace; namespace.MyClass.staticHelper(new namespace.MyClass()); };

Avoid accessing properties of an aliased type, unless it is an enum.

/** @enum {string} */ some.long.namespace.Fruit = { APPLE: 'a', BANANA: 'b' }; myapp.main = function() { var Fruit = some.long.namespace.Fruit; switch (fruit) { case Fruit.APPLE: ... case Fruit.BANANA: ... } }; myapp.main = function() { var MyClass = some.long.namespace.MyClass; MyClass.staticHelper(null); };

Never create aliases in the global scope. Use them only in function blocks.

Filenames should be all lowercase in order to avoid confusion on case-sensitive platforms. Filenames should end in .js, and should contain no punctuation except for - or _ (prefer - to _).

Must always succeed without side effects.

You can control how your objects string-ify themselves by defining a custom toString() method. This is fine, but you need to ensure that your method (1) always succeeds and (2) does not have side-effects. If your method doesn't meet these criteria, it's very easy to run into serious problems. For example, if toString() calls a method that does an assert, assert might try to output the name of the object in which it failed, which of course requires calling toString().

OK

It isn't always possible to initialize variables at the point of declaration, so deferred initialization is fine.

Always

Always use explicit scope - doing so increases portability and clarity. For example, don't rely on window being in the scope chain. You might want to use your function in another application for which window is not the content window.

Expand for more info.

We follow the C++ formatting rules in spirit, with the following additional clarifications.

Because of implicit semicolon insertion, always start your curly braces on the same line as whatever they're opening. For example:

if (something) { // ... } else { // ... }

Single-line array and object initializers are allowed when they fit on a line:

var arr = [1, 2, 3]; // No space after [ or before ]. var obj = {a: 1, b: 2, c: 3}; // No space after { or before }.

Multiline array initializers and object initializers are indented 2 spaces, just like blocks.

// Object initializer. var inset = { top: 10, right: 20, bottom: 15, left: 12 }; // Array initializer. this.rows_ = [ '"Slartibartfast" <fjordmaster@magrathea.com>', '"Zaphod Beeblebrox" <theprez@universe.gov>', '"Ford Prefect" <ford@theguide.com>', '"Arthur Dent" <has.no.tea@gmail.com>', '"Marvin the Paranoid Android" <marv@googlemail.com>', 'the.mice@magrathea.com' ]; // Used in a method call. goog.dom.createDom(goog.dom.TagName.DIV, { id: 'foo', className: 'some-css-class', style: 'display:none' }, 'Hello, world!');

Long identifiers or values present problems for aligned initialization lists, so always prefer non-aligned initialization. For example:

CORRECT_Object.prototype = { a: 0, b: 1, lengthyName: 2 };

Not like this:

WRONG_Object.prototype = { a : 0, b : 1, lengthyName: 2 };

When possible, all function arguments should be listed on the same line. If doing so would exceed the 80-column limit, the arguments must be line-wrapped in a readable way. To save space, you may wrap as close to 80 as possible, or put each argument on its own line to enhance readability. The indentation may be either four spaces, or aligned to the parenthesis. Below are the most common patterns for argument wrapping:

// Four-space, wrap at 80. Works with very long function names, survives // renaming without reindenting, low on space. goog.foo.bar.doThingThatIsVeryDifficultToExplain = function( veryDescriptiveArgumentNumberOne, veryDescriptiveArgumentTwo, tableModelEventHandlerProxy, artichokeDescriptorAdapterIterator) { // ... }; // Four-space, one argument per line. Works with long function names, // survives renaming, and emphasizes each argument. goog.foo.bar.doThingThatIsVeryDifficultToExplain = function( veryDescriptiveArgumentNumberOne, veryDescriptiveArgumentTwo, tableModelEventHandlerProxy, artichokeDescriptorAdapterIterator) { // ... }; // Parenthesis-aligned indentation, wrap at 80. Visually groups arguments, // low on space. function foo(veryDescriptiveArgumentNumberOne, veryDescriptiveArgumentTwo, tableModelEventHandlerProxy, artichokeDescriptorAdapterIterator) { // ... } // Parenthesis-aligned, one argument per line. Visually groups and // emphasizes each individual argument. function bar(veryDescriptiveArgumentNumberOne, veryDescriptiveArgumentTwo, tableModelEventHandlerProxy, artichokeDescriptorAdapterIterator) { // ... }

When declaring an anonymous function in the list of arguments for a function call, the body of the function is indented two spaces from the left edge of the function call or the statement, not two spaces from the left edge of the function keyword. This is to make the body of the anonymous function easier to read (i.e. not be all squished up into the right half of the screen).

var names = items.map(function(item) { return item.name; }); prefix.something.reallyLongFunctionName('whatever', function(a1, a2) { if (a1.equals(a2)) { someOtherLongFunctionName(a1); } else { andNowForSomethingCompletelyDifferent(a2.parrot); } });

In fact, except for array and object initializers , and passing anonymous functions, all wrapped lines should be indented either left-aligned to the expression above, or indented four spaces, not indented two spaces.

someWonderfulHtml = '
' + getEvenMoreHtml(someReallyInterestingValues, moreValues, evenMoreParams, 'a duck', true, 72, slightlyMoreMonkeys(0xfff)) + '
'; thisIsAVeryLongVariableName = hereIsAnEvenLongerOtherFunctionNameThatWillNotFitOnPrevLine(); thisIsAVeryLongVariableName = 'expressionPartOne' + someMethodThatIsLong() + thisIsAnEvenLongerOtherFunctionNameThatCannotBeIndentedMore(); someValue = this.foo( shortArg, 'Some really long string arg - this is a pretty common case, actually.', shorty2, this.bar()); if (searchableCollection(allYourStuff).contains(theStuffYouWant) && !ambientNotification.isActive() && (client.isAmbientSupported() || client.alwaysTryAmbientAnyways())) { ambientNotification.activate(); }

Use newlines to group logically related pieces of code. For example:

doSomethingTo(x); doSomethingElseTo(x); andThen(x); nowDoSomethingWith(y); andNowWith(z);

Always put the operator on the preceding line, so that you don't have to think about implicit semi-colon insertion issues. Otherwise, line breaks and indentation follow the same rules as in other Google style guides.

var x = a ? b : c; // All on one line if it will fit. // Indentation +4 is OK. var y = a ? longButSimpleOperandB : longButSimpleOperandC; // Indenting to the line position of the first operand is also OK. var z = a ? moreComplicatedB : moreComplicatedC;
Only where required

Use sparingly and in general only where required by the syntax and semantics.

Never use parentheses for unary operators such as delete, typeof and void or after keywords such as return, throw as well as others (case, in or new).

Prefer ' over "

For consistency single-quotes (') are preferred to double-quotes ("). This is helpful when creating strings that include HTML:

var msg = 'This is some HTML';
Encouraged, use JSDoc annotations @private and @protected

We recommend the use of the JSDoc annotations @private and @protected to indicate visibility levels for classes, functions, and properties.

@private global variables and functions are only accessible to code in the same file.

Constructors marked @private may only be instantiated by code in the same file and by their static and instance members. @private constructors may also be accessed anywhere in the same file for their public static properties and by the instanceof operator.

Global variables, functions, and constructors should never be annotated @protected.

// File 1. // AA_PrivateClass_ and AA_init_ are accessible because they are global // and in the same file. /** * @private * @constructor */ AA_PrivateClass_ = function() { }; /** @private */ function AA_init_() { return new AA_PrivateClass_(); } AA_init_();

@private properties are accessible to all code in the same file, plus all static methods and instance methods of that class that "owns" the property, if the property belongs to a class. They cannot be accessed or overridden from a subclass in a different file.

@protected properties are accessible to all code in the same file, plus any static methods and instance methods of any subclass of a class that "owns" the property.

Note that these semantics differ from those of C++ and Java, in that they grant private and protected access to all code in the same file, not just in the same class or class hierarchy. Also, unlike in C++, private properties cannot be overriden by a subclass.

// File 1. /** @constructor */ AA_PublicClass = function() { }; /** @private */ AA_PublicClass.staticPrivateProp_ = 1; /** @private */ AA_PublicClass.prototype.privateProp_ = 2; /** @protected */ AA_PublicClass.staticProtectedProp = 31; /** @protected */ AA_PublicClass.prototype.protectedProp = 4; // File 2. /** * @return {number} The number of ducks we've arranged in a row. */ AA_PublicClass.prototype.method = function() { // Legal accesses of these two properties. return this.privateProp_ + AA_PublicClass.staticPrivateProp_; }; // File 3. /** * @constructor * @extends {AA_PublicClass} */ AA_SubClass = function() { // Legal access of a protected static property. AA_PublicClass.staticProtectedProp = this.method(); }; goog.inherits(AA_SubClass, AA_PublicClass); /** * @return {number} The number of ducks we've arranged in a row. */ AA_SubClass.prototype.method = function() { // Legal access of a protected instance property. return this.protectedProp; };
Encouraged and enforced by the compiler.

When documenting a type in JSDoc, be as specific and accurate as possible. The types we support are JS2 style types and JS1.x types.

The JS2 proposal contained a language for specifying JavaScript types. We use this language in JsDoc to express the types of function parameters and return values.

As the JS2 proposal has evolved, this language has changed. The compiler still supports old syntaxes for types, but those syntaxes are deprecated.

Operator Name Syntax Description Deprecated Syntaxes
Type Name {boolean}, {Window}, {goog.ui.Menu} Simply the name of a type.
Type Application {Array.<string>}
An array of strings.

{Object.<string, number>}
An object in which the keys are strings and the values are numbers.

Patameterizes a type, by applying a set of type arguments to that type. The idea is analogous to generics in Java.
Type Union {(number|boolean)}
A number or a boolean.
Indicates that a value might have type A OR type B. {(number,boolean)}, {number|boolean}, {(number||boolean)}
Record Type {{myNum: number, myObject}}
An anonymous type with the given type members.

Indicates that the value has the specified members with the specified types. In this case, myNum with a type number and myObject with any type.

Notice that the braces are part of the type syntax. For example, to denote an Array of objects that have a length property, you might write Array.<{length}>.

Nullable type {?number}
A number or NULL.
Indicates that a value is type A or null. By default, all object types are nullable. NOTE: Function types are not nullable. {number?}
Non-nullable type {!Object}
An Object, but never the null value.
Indicates that a value is type A and not null. By default, all value types (boolean, number, string, and undefined) are not nullable. {Object!}
Function Type {function(string, boolean)}
A function that takes two arguments (a string and a boolean), and has an unknown return value.
Specifies a function.
Function Return Type {function(): number}
A function that takes no arguments and returns a number.
Specifies a function return type.
Function this Type {function(this:goog.ui.Menu, string)}
A function that takes one argument (a string), and executes in the context of a goog.ui.Menu.
Specifies the context type of a function type.
Variable arguments {function(string, ...[number]): number}
A function that takes one argument (a string), and then a variable number of arguments that must be numbers.
Specifies variable arguments to a function.
Variable arguments (in @param annotations) @param {...number} var_args
A variable number of arguments to an annotated function.
Specifies that the annotated function accepts a variable number of arguments.
Function optional arguments {function(?string=, number=)}
A function that takes one optional, nullable string and one optional number as arguments. The = syntax is only for function type declarations.
Specifies optional arguments to a function.
Function optional arguments (in @param annotations) @param {number=} opt_argument
An optional parameter of type number.
Specifies that the annotated function accepts an optional argument.
The ALL type {*} Indicates that the variable can take on any type.

Type Example Value Examples Description
number 1 1.0 -5 1e5 Math.PI
Number new Number(true) Number object
string 'Hello' "World" String(42) String value
String new String('Hello') new String(42) String object
boolean true false Boolean(0) Boolean value
Boolean new Boolean(true) Boolean object
RegExp new RegExp('hello') /world/g
Date new Date new Date()
null null
undefined undefined
void function f() { return; } No return value
Array ['foo', 0.3, null] [] Untyped Array
Array.<number> [11, 22, 33] An Array of numbers
Array.<Array.<string>> [['one', 'two', 'three'], ['foo', 'bar']] Array of Arrays of strings
Object {} {foo: 'abc', bar: 123, baz: null}
Object.<string> {'foo': 'bar'} An Object in which the values are strings.
Object.<number, string> var obj = {}; obj[1] = 'bar'; An Object in which the keys are numbers and the values are strings.

Note that in JavaScript, the keys are always implicitly coverted to strings, so obj['1'] == obj[1]. So the key wil always be a string in for...in loops. But the compiler will verify the type if the key when indexing into the object.

Function function(x, y) { return x * y; } Function object
function(number, number): number function(x, y) { return x * y; } function value
SomeClass /** @constructor */ function SomeClass() {} new SomeClass();
SomeInterface /** @interface */ function SomeInterface() {} SomeInterface.prototype.draw = function() {};
project.MyClass /** @constructor */ project.MyClass = function () {} new project.MyClass()
project.MyEnum /** @enum {string} */ project.MyEnum = { BLUE: '#0000dd', RED: '#dd0000' }; Enumeration
Element document.createElement('div') Elements in the DOM.
Node document.body.firstChild Nodes in the DOM.
HTMLInputElement htmlDocument.getElementsByTagName('input')[0] A specific type of DOM element.

Because JavaScript is a loosely-typed language, it is very important to understand the subtle differences between optional, nullable, and undefined function parameters and class properties.

Object types (also known as reference types) are nullable by default. NOTE: Function types are not nullable by default. An object is defined as anything except a string, number, boolean, undefined, or null. For example, the following declaration

/** * Some class, initialized with a value. * @param {Object} value Some value. * @constructor */ function MyClass(value) { /** * Some value. * @type {Object} * @private */ this.myValue_ = value; }

tells the compiler that the myValue_ property holds either an Object or null. If myValue_ must never be null, it should be declared like this:

/** * Some class, initialized with a non-null value. * @param {!Object} value Some value. * @constructor */ function MyClass(value) { /** * Some value. * @type {!Object} * @private */ this.myValue_ = value; }

This way, if the compiler can determine that somewhere in the code MyClass is initialized with a null value, it will issue a warning.

Optional parameters to functions may be undefined at runtime, so if they are assigned to class properties, those properties must be declared accordingly:

/** * Some class, initialized with an optional value. * @param {Object=} opt_value Some value (optional). * @constructor */ function MyClass(opt_value) { /** * Some value. * @type {Object|undefined} * @private */ this.myValue_ = opt_value; }

This tells the compiler that myValue_ may hold an Object, null, or remain undefined.

Note that the optional parameter opt_value is declared to be of type {Object=}, not {Object|undefined}. This is because optional parameters may, by definition, be undefined. While there is no harm in explicitly declaring an optional parameter as possibly undefined, it is both unnecessary and makes the code harder to read.

Finally, note that being nullable and being optional are orthogonal properties. The following four declarations are all different:

/** * Takes four arguments, two of which are nullable, and two of which are * optional. * @param {!Object} nonNull Mandatory (must not be undefined), must not be null. * @param {Object} mayBeNull Mandatory (must not be undefined), may be null. * @param {!Object=} opt_nonNull Optional (may be undefined), but if present, * must not be null! * @param {Object=} opt_mayBeNull Optional (may be undefined), may be null. */ function strangeButTrue(nonNull, mayBeNull, opt_nonNull, opt_mayBeNull) { // ... };
Use JSDoc

We use JSDoc comments to document files, classes, methods and properties. Inline comments should be of the // variety. Additionally, we follow the C++ style for comments in spirit. This means you should have:

Avoid sentence fragments. Start sentences with a properly capitalized word, and end them with punctuation.

Pretend there's some novice programmer that's going to come along and have to maintain the code after you. There very well just might be!

There are now many compiler passes that extract type information from JSDoc, in order to provide better code validation, removal, and compression. It is, therefore, very important that you use full and correct JSDoc.

The top level comment is designed to orient readers unfamiliar with the code to what is in this file. It should provide a description of the file's contents, its author(s), and any dependencies or compatibility information. As an example:

// Copyright 2009 Google Inc. All Rights Reserved. /** * @fileoverview Description of file, its uses and information * about its dependencies. * @author user@google.com (Firstname Lastname) */

Classes must be documented with a description and usage. The constructor parameters must also be documented. If the class inherits from another class, that should be documented with an @extends tag. If the class implements an interface, that should be documented with an @implements tag.

/** * Class making something fun and easy. * @param {string} arg1 An argument that makes this more interesting. * @param {Array.<number>} arg2 List of numbers to be processed. * @constructor * @extends {goog.Disposable} */ project.MyClass = function(arg1, arg2) { // ... }; goog.inherits(project.MyClass, goog.Disposable);

A description must be provided along with parameters. Use full sentences. Method descriptions should start with a sentence written in the third person declarative voice.

/** * Converts text to some completely different text. * @param {string} arg1 An argument that makes this more interesting. * @return {string} Some return value. */ project.MyClass.prototype.someMethod = function(arg1) { // ... }; /** * Operates on an instance of MyClass and returns something. * @param {project.MyClass} obj Instance of MyClass which leads to a long * comment that needs to be wrapped to two lines. * @return {boolean} Whether something occured. */ function PR_someMethod(obj) { // ... }

For simple getters that take no parameters, the description can be omitted.

/** * @return {Element} The element for the component. */ goog.ui.Component.prototype.getElement = function() { return this.element_; };

It is also nice to have comments for properties.

/** * Maximum number of things per pane. * @type {number} */ project.MyClass.prototype.someProperty = 4;

In cases where type-checking doesn't accurately infer the type of an expression, it is possible to add a type cast comment by adding a type annotation comment and enclosing the expression in parenthesis. The parentheses are required, and may surround the type annotation comment as well.

/** @type {number} */ (x) (/** @type {number} */ x)

If you have to line break a @param, @return, @supported, @this or @deprecated you should treat this as breaking a code statement and indent it four spaces.

/** * Illustrates line wrapping for long param/return descriptions. * @param {string} foo This is a param with a description too long to fit in * one line. * @return {number} This returns something that has a description too long to * fit in one line. */ project.MyClass.prototype.method = function(foo) { return 5; };

You should not indent the @fileoverview command.

Even though it is not preferred, it is also acceptable to line up the description. This has the side effect that you will have to realign the text every time you change a variable name so this will soon get your code out of sync.

/** * This is NOT the preferred indentation method. * @param {string} foo This is a param with a description too long to fit in * one line. * @return {number} This returns something that has a description too long to * fit in one line. */ project.MyClass.prototype.method = function(foo) { return 5; };
/** * Enum for tri-state values. * @enum {number} */ project.TriState = { TRUE: 1, FALSE: -1, MAYBE: 0 };

Note that enums are also valid types and thus can be used as parameter types, etc.

/** * Sets project state. * @param {project.TriState} state New project state. */ project.setState = function(state) { // ... };

Sometimes types can get complicated. A function that accepts content for an Element might look like:

/** * @param {string} tagName * @param {(string|Element|Text|Array.<Element>|Array.<Text>)} contents * @return {Element} */ goog.createElement = function(tagName, contents) { ... };

You can define commonly used type expressions with a @typedef tag. For example,

/** @typedef {(string|Element|Text|Array.<Element>|Array.<Text>)} */ goog.ElementContent; /** * @param {string} tagName * @param {goog.ElementContent} contents * @return {Element} */ goog.createElement = function(tagName, contents) { ... };
Tag Template & Examples Description Type-Checking Support
@param @param {Type} varname Description

For example:

/** * Queries a Baz for items. * @param {number} groupNum Subgroup id to query. * @param {string|number|null} term An itemName, * or itemId, or null to search everything. */ goog.Baz.prototype.query = function(groupNum, term) { // ... };
Used with method, function and constructor calls to document the arguments of a function. Fully supported.
@return @return {Type} Description

For example:

/** * @return {string} The hex ID of the last item. */ goog.Baz.prototype.getLastId = function() { // ... return id; };
Used with method and function calls to document the return type. When writing descriptions for boolean parameters, prefer "Whether the component is visible" to "True if the component is visible, false otherwise". If there is no return value, do not use an @return tag. Fully supported.
@author @author username@google.com (first last)

For example:

/** * @fileoverview Utilities for handling textareas. * @author kuth@google.com (Uthur Pendragon) */
Document the author of a file or the owner of a test, generally only used in the @fileoverview comment. Unrelated to type checking.
@see @see Link

For example:

/** * Adds a single item, recklessly. * @see #addSafely * @see goog.Collect * @see goog.RecklessAdder#add ...
Reference a lookup to another class function or method. Unrelated to type checking.
@fileoverview @fileoverview Description

For example:

/** * @fileoverview Utilities for doing things that require this very long * but not indented comment. * @author kuth@google.com (Uthur Pendragon) */
Makes the comment block provide file level information. Unrelated to type checking.
@constructor @constructor

For example:

/** * A rectangle. * @constructor */ function GM_Rect() { ... }
Used in a class's documentation to indicate the constructor. Yes. If omitted the compiler will prohibit instantiation.
@interface @interface

For example:

/** * A shape. * @interface */ function Shape() {}; Shape.prototype.draw = function() {}; /** * A polygon. * @interface * @extends {Shape} */ function Polygon() {}; Polygon.prototype.getSides = function() {};
Used to indicate that the function defines an inteface. Yes. The compiler will warn about instantiating an interface.
@type @type Type
@type {Type}

For example:

/** * The message hex ID. * @type {string} */ var hexId = hexId;
Identifies the type of a variable, property, or expression. Curly braces are not required around most types, but some projects mandate them for all types, for consistency. Yes
@extends @extends Type
@extends {Type}

For example:

/** * Immutable empty node list. * @constructor * @extends goog.ds.BasicNodeList */ goog.ds.EmptyNodeList = function() { ... };
Used with @constructor to indicate that a class inherits from another class. Curly braces around the type are optional. Yes
@implements @implements Type
@implements {Type}

For example:

/** * A shape. * @interface */ function Shape() {}; Shape.prototype.draw = function() {}; /** * @constructor * @implements {Shape} */ function Square() {}; Square.prototype.draw = function() { ... };
Used with @constructor to indicate that a class implements an interface. Curly braces around the type are optional. Yes. The compiler will warn about incomplete implementations of interfaces.
@lends @lends objectName
@lends {objectName}

For example:

goog.object.extend( Button.prototype, /** @lends {Button.prototype} */ { isButton: function() { return true; } });
Indicates that the keys of an object literal should be treated as properties of some other object. This annotation should only appear on object literals.

Notice that the name in braces is not a type name like in other annotations. It's an object name. It names the object on which the properties are "lent". For example, @type {Foo} means "an instance of Foo", but @lends {Foo} means "the constructor Foo".

The JSDoc Toolkit docs have more information on this annotation.

Yes
@private @private

For example:

/** * Handlers that are listening to this logger. * @type Array.<Function> * @private */ this.handlers_ = [];
Used in conjunction with a trailing underscore on the method or property name to indicate that the member is private. Trailing underscores may eventually be deprecated as tools are updated to enforce @private. Enforced with a flag.
@protected @protected

For example:

/** * Sets the component's root element to the given element. Considered * protected and final. * @param {Element} element Root element for the component. * @protected */ goog.ui.Component.prototype.setElementInternal = function(element) { // ... };
Used to indicate that the member or property is protected. Should be used in conjunction with names with no trailing underscore. Enforced with a flag.
@this @this Type
@this {Type}

For example:

pinto.chat.RosterWidget.extern('getRosterElement', /** * Returns the roster widget element. * @this pinto.chat.RosterWidget * @return {Element} */ function() { return this.getWrappedComponent_().getElement(); });
The type of the object in whose context a particular method is called. Required when the this keyword is referenced from a function that is not a prototype method. Yes
@supported @supported Description

For example:

/** * @fileoverview Event Manager * Provides an abstracted interface to the * browsers' event systems. * @supported So far tested in IE6 and FF1.5 */
Used in a fileoverview to indicate what browsers are supported by the file. Unrelated to type checking.
@enum @enum {Type}

For example:

/** * Enum for tri-state values. * @enum {number} */ project.TriState = { TRUE: 1, FALSE: -1, MAYBE: 0 };
Used for documenting enum types. Fully supported. If Type is omitted, number assumed.
@deprecated @deprecated Description

For example:

/** * Determines whether a node is a field. * @return {boolean} True if the contents of * the element are editable, but the element * itself is not. * @deprecated Use isField(). */ BN_EditUtil.isTopEditableField = function(node) { // ... };
Used to tell that a function, method or property should not be used any more. Always provide instructions on what callers should use instead. Unrelated to type checking
@override @override

For example:

/** * @return {string} Human-readable representation of project.SubClass. * @override */ project.SubClass.prototype.toString() { // ... };
Indicates that a method or property of a subclass intentionally hides a method or property of the superclass. If no other documentation is included, the method or property also inherits documentation from its superclass. Yes
@inheritDoc @inheritDoc

For example:

/** @inheritDoc */ project.SubClass.prototype.toString() { // ... };
Indicates that a method or property of a subclass intentionally hides a method or property of the superclass, and has exactly the same documentation. Notice that @inheritDoc implies @override. Yes
@code {@code ...}

For example:

/** * Moves to the next position in the selection. * Throws {@code goog.iter.StopIteration} when it * passes the end of the range. * @return {Node} The node at the next position. */ goog.dom.RangeIterator.prototype.next = function() { // ... };
Indicates that a term in a JSDoc description is code so it may be correctly formatted in generated documentation. Not applicable.
@license or @preserve @license Description

For example:

/** * @preserve Copyright 2009 SomeThirdParty. * Here is the full license text and copyright * notice for this file. Note that the notice can span several * lines and is only terminated by the closing star and slash: */
Anything marked by @license or @preserve will be retained by the compiler and output at the top of the compiled code for that file. This annotation allows important notices (such as legal licenses or copyright text) to survive compilation unchanged. Line breaks are preserved. Unrelated to type checking.
@noalias @noalias

For example:

/** @noalias */ function Range() {}
Used in an externs file to indicate to the compiler that the variable or function should not be aliased as part of the alias externals pass of the compiler. Unrelated to type checking.
@define @define {Type} description

For example:

/** @define {boolean} */ var TR_FLAGS_ENABLE_DEBUG = true; /** @define {boolean} */ goog.userAgent.ASSUME_IE = false;
Indicates a constant that can be overridden by the compiler at compile-time. In the example, the compiler flag --define='goog.userAgent.ASSUME_IE=true' could be specified in the BUILD file to indicate that the constant goog.userAgent.ASSUME_IE should be replaced with true. Unrelated to type checking.
@export @export

For example:

/** @export */ foo.MyPublicClass.prototype.myPublicMethod = function() { // ... };

Given the code on the left, when the compiler is run with the --generate_exports flag, it will generate the code:

goog.exportSymbol('foo.MyPublicClass.prototype.myPublicMethod', foo.MyPublicClass.prototype.myPublicMethod);

which will export the symbols to uncompiled code. Code that uses the @export annotation must either

  1. include //javascript/closure/base.js, or
  2. define both goog.exportSymbol and goog.exportProperty with the same method signature in their own codebase.
Unrelated to type checking.
@const @const

For example:

/** @const */ var MY_BEER = 'stout'; /** * My namespace's favorite kind of beer. * @const * @type {string} */ mynamespace.MY_BEER = 'stout'; /** @const */ MyClass.MY_BEER = 'stout';

Marks a variable as read-only and suitable for inlining. Generates warnings if it is rewritten.

Constants should also be ALL_CAPS, but the annotation should help eliminate reliance on the naming convention. Although @final is listed at jsdoc.org and is supported as equivalent to @const in the compiler, it is discouraged. @const is consistent with JS1.5's const keyword. Note that changes to properties of const objects are not currently prohibited by the compiler (inconsistent with C++ const semantics). The type declaration can be omitted if it can be clearly inferred. If present, it must be on its own line. An additional comment about the variable is optional.

Supported by type checking.
@nosideeffects @nosideeffects

For example:

/** @nosideeffects */ function noSideEffectsFn1() { // ... }; /** @nosideeffects */ var noSideEffectsFn2 = function() { // ... }; /** @nosideeffects */ a.prototype.noSideEffectsFn3 = function() { // ... };
This annotation can be used as part of function and constructor declarations to indicate that calls to the declared function have no side-effects. This annotation allows the compiler to remove calls to these functions if the return value is not used. Unrelated to type checking.
@typedef @typedef

For example:

/** @typedef {(string|number)} */ goog.NumberLike; /** @param {goog.NumberLike} x A number or a string. */ goog.readNumber = function(x) { ... }
This annotation can be used to declare an alias of a more complex type. Yes
@externs @externs

For example:

/** * @fileoverview This is an externs file. * @externs */ var document;

Declares an externs file.

No

You may also see other types of JSDoc annotations in third-party code. These annotations appear in the JSDoc Toolkit Tag Reference but are currently discouraged in Google code. You should consider them "reserved" names for future use. These include:

  • @augments
  • @argument
  • @borrows
  • @class
  • @constant
  • @constructs
  • @default
  • @event
  • @example
  • @field
  • @function
  • @ignore
  • @inner
  • @link
  • @memberOf
  • @name
  • @namespace
  • @property
  • @public
  • @requires
  • @returns
  • @since
  • @static
  • @version

Like JavaDoc, JSDoc supports many HTML tags, like <code>, <pre>, <tt>, <strong>, <ul>, <ol>, <li>, <a>, and others.

This means that plaintext formatting is not respected. So, don't rely on whitespace to format JSDoc:

/** * Computes weight based on three factors: * items sent * items received * last timestamp */

It'll come out like this:

Computes weight based on three factors: items sent items received items received

Instead, do this:

/** * Computes weight based on three factors: * <ul> * <li>items sent * <li>items received * <li>last timestamp * </ul> */

Also, don't include HTML or HTML-like tags unless you want them to be interpreted as HTML.

/** * Changes <b> tags to <span> tags. */

It'll come out like this:

Changes tags to tags.

On the other hand, people need to be able to read this in its plaintext form too, so don't go overboard with the HTML:

/** * Changes &lt;b&gt; tags to &lt;span&gt; tags. */

People will know what you're talking about if you leave the angle-brackets out, so do this:

/** * Changes 'b' tags to 'span' tags. */
Encouraged

Use of JS compilers such as the Closure Compiler is encouraged.

JavaScript tidbits

The following are all false in boolean expressions:

  • null
  • undefined
  • '' the empty string
  • 0 the number

But be careful, because these are all true:

  • '0' the string
  • [] the empty array
  • {} the empty object

This means that instead of this:

while (x != null) {

you can write this shorter code (as long as you don't expect x to be 0, or the empty string, or false):

while (x) {

And if you want to check a string to see if it is null or empty, you could do this:

if (y != null && y != '') {

But this is shorter and nicer:

if (y) {

Caution: There are many unintuitive things about boolean expressions. Here are some of them:

  • Boolean('0') == true
    '0' != true
  • 0 != null
    0 == []
    0 == false
  • Boolean(null) == false
    null != true
    null != false
  • Boolean(undefined) == false
    undefined != true
    undefined != false
  • Boolean([]) == true
    [] != true
    [] == false
  • Boolean({}) == true
    {} != true
    {} != false

Instead of this:

if (val != 0) { return foo(); } else { return bar(); }

you can write this:

return val ? foo() : bar();

The ternary conditional is also useful when generating HTML:

var html = '<input type="checkbox"' + (isChecked ? ' checked' : '') + (isEnabled ? '' : ' disabled') + ' name="foo">';

These binary boolean operators are short-circuited, and evaluate to the last evaluated term.

"||" has been called the 'default' operator, because instead of writing this:

/** @param {*=} opt_win */ function foo(opt_win) { var win; if (opt_win) { win = opt_win; } else { win = window; } // ... }

you can write this:

/** @param {*=} opt_win */ function foo(opt_win) { var win = opt_win || window; // ... }

"&&" is also useful for shortening code. For instance, instead of this:

if (node) { if (node.kids) { if (node.kids[index]) { foo(node.kids[index]); } } }

you could do this:

if (node && node.kids && node.kids[index]) { foo(node.kids[index]); }

or this:

var kid = node && node.kids && node.kids[index]; if (kid) { foo(kid); }

However, this is going a little too far:

node && node.kids && node.kids[index] && foo(node.kids[index]);

It is common to see this:

function listHtml(items) { var html = '<div class="foo">'; for (var i = 0; i < items.length; ++i) { if (i > 0) { html += ', '; } html += itemHtml(items[i]); } html += '</div>'; return html; }

but this is slow in Internet Explorer, so it is better to do this:

function listHtml(items) { var html = []; for (var i = 0; i < items.length; ++i) { html[i] = itemHtml(items[i]); } return '<div class="foo">' + html.join(', ') + '</div>'; }

You can also use an array as a stringbuilder, and convert it into a string with myArray.join(''). Note that since assigning values to an array is faster than using push() you should use assignment where possible.

Node lists are often implemented as node iterators with a filter. This means that getting a property like length is O(n), and iterating over the list by re-checking the length will be O(n^2).

var paragraphs = document.getElementsByTagName('p'); for (var i = 0; i < paragraphs.length; i++) { doSomething(paragraphs[i]); }

It is better to do this instead:

var paragraphs = document.getElementsByTagName('p'); for (var i = 0, paragraph; paragraph = paragraphs[i]; i++) { doSomething(paragraph); }

This works well for all collections and arrays as long as the array does not contain things that are treated as boolean false.

In cases where you are iterating over the childNodes you can also use the firstChild and nextSibling properties.

var parentNode = document.getElementById('foo'); for (var child = parentNode.firstChild; child; child = child.nextSibling) { doSomething(child); }

BE CONSISTENT.

If you're editing code, take a few minutes to look at the code around you and determine its style. If they use spaces around all their arithmetic operators, you should too. If their comments have little boxes of hash marks around them, make your comments have little boxes of hash marks around them too.

The point of having style guidelines is to have a common vocabulary of coding so people can concentrate on what you're saying rather than on how you're saying it. We present global style rules here so people know the vocabulary, but local style is also important. If code you add to a file looks drastically different from the existing code around it, it throws readers out of their rhythm when they go to read it. Avoid this.

Revision 2.9

Aaron Whyte
Bob Jervis
Dan Pupius
Erik Arvidsson
Fritz Schneider
Robby Walker