Head First into Unity with UnityScript

From Unify Community Wiki
Jump to: navigation, search

Written by Tonio Loewald (a.k.a. podperson)

Also see this article which outlines the technical differences: UnityScript_versus_JavaScript (it's a helpful article, but in places outdated or wrong).

Note: this tutorial assumes you know the basics of JavaScript and programming. We're not going to go over all that stuff from scratch. If you don't know anything at all about programming, well ... you should probably get a book oriented at learning to program (DON'T learn JavaScript expecting it to help you program in UnityScript — it's more of a cross between JavaScript and Java or C#. Indeed, if you don't want to learn JavaScript you're probably better off learning C#), or take a class (or perhaps check this thread). The fundamentals of programming are a major topic in and of themselves.

Contents

Unity's "JavaScript" vs. the JavaScript you probably know

Use #pragma strict

#pragma strict

It's a good habit to get into and required for iOS development. #pragma strict will enforce stricter type checking, generate more useful error messages sooner, and will encourage good programming habits.

Use enum

enum WeaponType { pistol, rifle, launcher}
var type : WeaponType = WeaponType.pistol;

This is a much more economical and less error-prone method than using strings.

It's different

Although Unity's JavaScript tries to be like the ECMAScript standard at least somewhat, it varies in many ways from other implementations of JavaScript that are based on the same standard. It is perhaps most similar to Microsoft's JScript, especially in that both are .NET languages. However, Unity's version was developed independently and there are a number of differences between the two.

It's fast

Unity JavaScript is compiled (and fast, which is excellent) but not so dynamic as JavaScript in browsers (which is interpreted).

Seriously, there is no speed difference between JavaScript, C#, and Boo (in Unity, that is). There are other pros and cons to each, but not speed.

Note: if you've been following recent developments in the browser wars you'll know that it's not strictly true that browser JavaScript is interpreted. JavaScript is in fact JIT compiled on all modern browsers, and rather impressively fast on some (FireFox, Safari, and Opera). Even so, it doesn't support strict typing (even as an option) which is very costly for performance (every time you perform arithmetic operations you need to check what your operands are and convert them if necessary). If the ECMAScript standard is modified (as Adobe has lobbied for) to allow for explicit type declarations, it's possible that appropriately written JavaScript will be able to execute an order of magnitude faster.

Even so, as things stand, "real" JavaScript is still about two orders of magnitude slower than Unity's UnityScript, even on Squirrelfish Extreme.

The var keyword is required

In JavaScript, if you don't use the var keyword when defining a variable, that variable is scoped as a global one.

function DoSomeStuff() {
   x = 3;
}
 
DoSomeStuff();
 
alert(x); // returns 3 ... in JavaScript (not in Unity's UnityScript)

To avoid this ambiguity for long-time JS users, the var keyword is required when defining variables, and that automatically restricts the variable's scope to the current scope.

function DoSomeStuff() {
   var x = 3;
}
 
DoSomeStuff();
 
print(x); // raises an error because x is not global in any sense.

UnityScript is classical, not prototypal

There's no messing around with .prototype in UnityScript. To make a class, you simply do this:

// Foo.js
var foo = "hello, world";
 
function doEet () {
  // does nothing, intended to be overridden
}

The compiler essentially adds some padding code to this (to make it a class definition) before it gets compiled. It effectively becomes something like:

// Foo.js
import UnityEngine;
class Foo extends MonoBehaviour {
  public var foo = "hello, world";
 
  public function doEet () {
    // does nothing, intended to be overridden
  }
}

Note the name of the class comes from the name of the file.

Here's a subclass:

// PrintingFoo.js
class PrintingFoo extends Foo {
   function doEet() {
    print( foo );
  }
}


Virtual functions can be used to override functions

In Unity's javascript you can create virtual functions.

class Foo
{
     virtual function DoSomething () 
     {
         Debug.Log("from base class");
     }
}
 
//SubFoo.js
class SubFoo extends Foo
{
     virtual function DoSomething() 
     {
          Debug.Log("from sub class");
     }
}
 
//Elsewhere
var foo : Foo = new SubFoo();
foo.DoSomething();//prints from sub class

If you want to call the parent class's method then you can use the keyword super. We could get SubFoo to call the function in Foo by writing:

class SubFoo extends Foo
{
     virtual function DoSomething()
     {
          super.DoSomething();
          Debug.Log("from sub class");
     }
}
 
//Elsewhere
var foo : Foo = new SubFoo();
foo.DoSomething();//prints "from base class" and "from sub class"

Consider Writing Mixins and Helpers Instead of Subclassing

It's very easy to write classes that know about one another and cooperate, and this is probably a better and more maintainable way of specializing objects in Unity than subclassing.

E.g.

/* Foo.js */
var bar : Bar;
 
function Start(){
  bar = gameObject.GetComponent(Bar);
}
 
function doEet(){
  // do my own thing
  if( bar ){
    bar.doEet();
  }
}
 
/* Bar.js */
  function doEet(){
    // do something special
  }

It's String (Mono's String class) and not string

var x : String;

Most of the string functionality you know and love from JavaScript is still there, but it's differently capitalized.


A note on how to "Split" Strings

You have to use the following syntax for splitting with characters :

var qualifiedName = "System.Integer myInt";
 
var name = qualifiedName.Split(" "[0]);

name[1] will contain "myInt" after the split.


For a list of available string functions, see Mono's documentation on the String class (http://go-mono.com/docs/monodoc.ashx?link=T%3aSystem.String%2f*)

You Must Declare Variables Before Using Them

You must declare variables before using them.

a = "fred"; // works in JavaScript (a is treated as a global), error in Unity
var a = "fred"; // a is now a string variable containing 'fred'
var b: String; // b is now a string variable, with no assigned value
b = "wilma";
var c; // c is now a dynamically typed variable with no assigned value
c = "barney";
c = 17;

a) You can (and often should) explicitly scope variables as private, public, etc. If you leave out the scope, public is the default.

b) Unity will implicitly type a variable if you assign it a value when you declare it. So:

var a = "fred"; // a is now of type String
a = 5; // ERROR! -- a is a String
var b : String = "fred"; // redundant

But:

var a; // a is dynamically typed;
a = "fred"; // works
a = 5; // works

Method (and Class) Names are Generally Capitalized

Method (and class) names are generally capitalized, except when they aren't. (It's confusing.) Basically, Unity's JavaScript is living in a .NET naming convention world (where methods are CamelCase and properties are camelCase), but is also trying to be like JavaScript (which, like C, is strongly biased towards lowercase and camelCase for everything).

e.g. in JavaScript typeof("fred") == 'string', but in Unity the type you write var a: String;

Lots More Types, Two Kinds of Array, No Object Syntax Sugar

JavaScript has, in essence, three types: number, string, and Object (with functions and arrays in essence being Objects). Unity's JavaScript has many more types, including:

Objects, which are NOT interchangeable with arrays, or Arrays (which are somewhat like JavaScript's objects, but not dynamic):

var a = new Object(); // works
a.fred = "wilma"; // runtime exception!

native arrays (which are not associative or dynamic):

var a = [1, 2, 3];
a.Push(4); // ERROR -- won't work!

If you want to make an array of a specific type:

public var friendsOfCarlotta : Transform[];

UnityScript Arrays (which are associative and dynamic):

var a = new Array();
a.Push(4); // This works

You can convert UnityScript Arrays to native arrays (which are faster but less flexible) using ToBuiltIn(ArrayType), e.g.:

var a = new Array();
a.Push(1);
a.Push(3.1415926535);
a.Push(17);
var b = a.ToBuiltin(float);

Integer types (including int, uint32, etc.):

Unity supports a large number of special case variations of the integer type. In general you'll know if you need them and don't need to worry otherwise.

Unity has many built-in classes (e.g. Vector3):

You'll become familiar with these as you work with Unity, in much the same way you become familiar with the DOM classes as you work with browsers. (Except Unity's classes suck much less than the DOM.)

One of the things that makes Unity a lot of fun to work with is that it uses a very liberal "mixin" strategy for classes. Usually you can "get at" the classes you need very quickly and easily. The most common example is the Transform class which quickly gets you to any associated class that's "attached" to the same object you're dealing with very quickly and easily.

E.g. in a typical behavior you'll have access to a variable named "transform" (referring to the instance of Transform associated with the object to which the behavior is attached). If you need the associated position (a Vector3) it's transform.position; if you need its GameObject it's transform.gameObject. If you need its renderer, it's transform.renderer. And so on. Generally, in Unity if you have one of an object's major properties you can quickly get at all its others.

a) Unity's String class lacks many of the nicer features of JavaScript's strings.

b) Unity's internal arrays are far less flexible than JavaScript's arrays or objects. However, various collection types such as List, Dictionary, Queue and so on are available. Internal arrays always have the advantage of the best speed.

Each .js File Implements A Class (By Default)

In general, sticking a bunch of variable and function definitions such as:

var x : int;
function y(){}

and saving it in Foo.js Is exactly equivalent to putting something like this in the same file:

class Foo extends MonoBehaviour {
  var x : int;
  function y(){}
}

But, you can also declare more than one class in the same file, which is particularly useful where you need utility classes which aren't descended from MonoBehaviour.

e.g.

class ButtonState {
  var currentState : int;
  var offset : Vector2;
}

If you do something like declare a subclass of MonoBehaviour in a file with a mismatched name (including different capitalization) you're asking for trouble.

It's important to understand that when you write a behavior script in JavaScript you are actually writing a class implementation, where:

a) The name of the class is the name of the script file (so if it's foo.js you can instance it elsewhere by saying var x = new foo()).

b) Certain "magic" method names will in fact implement event handlers (e.g. Start(), FixedUpdate() etc.). In any event, a function declaration is a method of the class you've written.

c) Code written outside function definitions inside your file are executing in the class's body. Variables declared in it are members of the class.

d) static functions and variables in a class are, in essence, class functions and variables.

This is all FAR more elegant than implementing classes in "real" JavaScript, but also somewhat restrictive ... mostly in a good way (you can't arbitrarily wire objects together the way you can in "real" JavaScript).

For example, if I create a new behavior and name it foo, the file will be named foo.js. Let's suppose foo.js looks like this:

public name : String; // when you drag the behavior onto a gameobject, these values will be visible and editable
public age : int; // other scripts which have a reference to this object (e.g. if they're attached to the same object) can see public functions
private favoriteColor : Color; // private members are NOT visible to other scripts, even if they have a reference to this object
public bestFriend : foo; // you can assign a value to bestFriend by dragging a gameObject with an attached copy of the foo behavior to this property. This will give you access to bestFriend's public methods and members
 
function Update(){
  // this function will be called every frame by Unity, so it's actually an event handler
  var t = transform; // transform is a property inherited from the gameObject the behavior is attached to
}
 
function Bar(){
  // this is just a function, if you don't call it yourself, it will never do anything
}

Calling Inherited Methods

super() is the inherited constructor, and super is "this" treated as a member of the superclass (so super.foo() calls the superclass's version of foo).

Semicolons Are Not Optional

Semicolons are generally optional in JavaScript (which has some ferocious logic to determine when a statement ends) but very much not optional in Unity.

var foo = 3 // OK in JavaScript but an error in Unity
foo += 17

Math is Mathf; Math.abs is Mathf.Abs

JavaScript's annoying Math library (because, really, stuff like abs() should just be in the language) becomes Unity's (also annoying) Mathf library. And remember, method names are usually capitalized (in Unity) so Math.abs() becomes Mathf.Abs().

Using Mono (.NET)

The UnityScript runtime environment makes extensive use of Mono (the open source clone of .NET). In fact, UnityScript is implemented in Boo, which is a language that runs on the Mono virtual machine and compiles to its native code. Many of a typical JavaScript runtime environment (e.g. String and Math libraries) are provided by Mono. You can tell when you're using Mono classes because their capitalization convention is CamelCase for class names (e.g. String) and methods (e.g. Mathf.Sin()) but camelCase for properties vs. JavaScript's which is camelCase for everything.

In order to use the Mono libraries it's most convenient to import them, e.g.

import System;
import System.IO;

Otherwise you need to specify the full namespace when calling functions, e.g. System.IO.File.Open instead of just File.Open. To find documentation on the libraries, the best place is probably http://msdn.microsoft.com/en-us/library/w0x726c2(v=VS.90).aspx The Mono documentation (http://docs.go-mono.com/) is technically more accurate for what's included with Unity, but is fairly incomplete at this time.

Mono Types

When a Mono function requires a char as an input, you can obtain one by simply indexing a string. E.g. if you wanted to pass the lowercase a as a char, you'd write:

"a"[0]

E.g. when using the String.Replace() function, you would use:

var s : String = "Whatever_it_may_be";
s = s.Replace("_"[0], " "[0]); // replace all the underscores with spaces

When working with Javascript Arrays (which are more flexible than builtins) you can easily convert them to the faster builtin arrays using ToBuiltIn(Type):

fastArray : SomeType[] = monoArray.ToBuiltin(SomeType);

UnityScript has access to generics (like in C#), so when using dynamically-sized arrays, you're better off using List instead of Javascript arrays. There's essentially no reason to use Javascript Arrays; Lists are faster and more functional in all cases. If you need an array of mixed types, you can use a List of Object, and then cast Object to the appropriate type when retrieving values. The syntax for generics in UnityScript is nearly identical to C#, except for an extra "." before the <> part. For example, "var myList = new List<int>();" in C# would become "var myList = new List.<int>();" in UnityScript.

Using 3rd Party .NET Libraries

3rd Party .NET libraries (e.g. XML-RPC.net) can be used by importing the .dll file as a new asset.

Debugging

Scripting errors will show in Unity as a "red x icon" in the window status bar. Click on the icon to bring up the console, showing a list of errors, which should be both informative and lead you to the line in the script that caused the problem.

Unity also generates useful warnings (with a "yellow ! icon") e.g. telling you a variable you've declared isn't used anywhere. Striving to write code that never generates warnings is a very useful habit.

The print() function will produce messages in the status bar and console, but only in MonoBehaviour classes. It's better to use Debug.Log("insert message here");. This does the same thing as print(), but everywhere. You can also use Debug.LogWarning and Debug.LogError to generate your own warning and error messages.

Debug.Break(); pauses the game in the editor at that exact point. It's very useful if you want to examine the state of objects when a particular situation occurs (e.g. you're trying to figure out what's going wrong when a particular object collides with another).

The editing GUI is completely live when running projects in the development environment (e.g. instances created at runtime appear in the browsers and you can click on them and look at their internal state).

Personal tools
Namespaces

Variants
Actions
Navigation
Extras
Toolbox