Basic tutorial

This step-by-step tutorial where we start with the setup of a Scala.js sbt project and end up having some user interaction and unit testing. The code created in this tutorial is available with one commit per step in the scalajs-tutorial repository on GitHub.

Step 0: Prerequisites

To go through this tutorial, you will need to download & install sbt (>= 0.13.0). Note that no prior sbt knowledge (only a working installation) is required to follow the tutorial.

Step 1: Setup

First create a new folder where your sbt project will go.

sbt Setup

To setup Scala.js in a new sbt project, we need to do two things:

  1. Add the Scala.js sbt plugin to the build
  2. Enable the plugin in the project

Adding the Scala.js sbt plugin is a one-liner in project/plugins.sbt (all file names we write in this tutorial are relative to the project root):

addSbtPlugin("org.scala-js" % "sbt-scalajs" % "0.6.5")

We also setup basic project settings and enable this plugin in the sbt build file (build.sbt, in the project root directory):

enablePlugins(ScalaJSPlugin)

name := "Scala.js Tutorial"

scalaVersion := "2.11.5" // or any other Scala version >= 2.10.2

Last, we need a project/build.properties to specify the sbt version (>= 0.13.7):

sbt.version=0.13.7

That is all we need to configure the build.

If at this point you prefer to use Eclipse or IDEA as your IDE, you may use sbteclipse to generate an Eclipse project, or import the sbt build from IDEA. Note that for compiling and running your application, you will still need to use sbt from the command line.

HelloWorld application

For starters, we add a very simple TutorialApp in the tutorial.webapp package. Create the file src/main/scala/tutorial/webapp/TutorialApp.scala:

package tutorial.webapp

import scala.scalajs.js.JSApp

object TutorialApp extends JSApp {
  def main(): Unit = {
    println("Hello world!")
  }
}

As you expect, this will simply print “HelloWorld” when run. To run this, simply launch sbt and invoke the run task:

$ sbt
> run
[info] Compiling 1 Scala source to (...)/scala-js-tutorial/target/scala-2.11/classes...
[info] Running tutorial.webapp.TutorialApp
Hello world!
[success] (...)

Troubleshooting: Should you experience errors with the PermGen size of the JVM at this point, you can increase it. Refer, for example, to this StackOverflow question.

Congratulations! You have successfully compiled and run your first Scala.js application. The code is actually run by a JavaScript interpreter. If you do not believe this (it happens to us occasionally), you can use the last command in sbt:

> last
(...)
[info] Running tutorial.webapp.TutorialApp
[debug] with JSEnv of type class org.scalajs.jsenv.rhino.RhinoJSEnv
[debug] with classpath of type class org.scalajs.core.tools.classpath.IRClasspath
[success] (...)

So your code has actually been executed by the Rhino JavaScript interpreter.

Rhino is cool because it runs out of the box, without having to install anything. But it is terribly slow. In general, you do not want to use it for your day-to-day development.

Node.js is a much more performant JavaScript engine. To run your code with Node.js, you need to install it, and enable FastOptStage in sbt using this command:

> set scalaJSStage in Global := FastOptStage
> run
[info] Fast optimizing C:\Users\Sepi\Documents\Projets\scalajs-tutorial\target\scala-2.11\scala-js-tutorial-fastopt.js
[info] Running tutorial.webapp.TutorialApp
Hello world!
[success] (...)

The last command now shows that this was run with Node.js:

> last
(...)
[info] Running tutorial.webapp.TutorialApp
[debug] with JSEnv of type class org.scalajs.jsenv.nodejs.NodeJSEnv
[debug] with classpath of type class org.scalajs.core.tools.classpath.LinkedClasspath
[success] (...)

The stage needs to be set once per sbt session. Alternatively, you can include the setting directly in your build.sbt, or, in order not to disturb your teammates, in a separate .sbt file (say, local.sbt):

scalaJSStage in Global := FastOptStage

This will enable the fastOpt stage by default when launching sbt.

Source maps in Node.js: To get your stack traces resolved on Node.js, you will have to install the source-map-support package.

npm install source-map-support

Step 2: Integrating with HTML

Now that we have a simple JavaScript application, we would like to use it in an HTML page. To do this, we need two steps:

  1. Generate a single JavaScript file out of our compiled code
  2. Create an HTML page which includes that file and calls the application

Generate JavaScript

To generate a single JavaScript file using sbt, just use the fastOptJS task:

> fastOptJS
[info] Fast optimizing (...)/scala-js-tutorial/target/scala-2.11/scala-js-tutorial-fastopt.js
[success] (...)

This will perform some fast optimizations and generate the target/scala-2.11/scala-js-tutorial-fastopt.js file containing the JavaScript code.

(It is possible that the [info] does not appear, if you have just run the program in fastOpt mode.)

Create the HTML Page

To load and launch the created JavaScript, you will need an HTML file. Create the file scalajs-tutorial-fastopt.html (or whatever name you prefer, for example index-dev.html) in the project root with the following content. We will go in the details right after.

<!DOCTYPE html>
<html>
  <head>
    <meta charset="UTF-8">
    <title>The Scala.js Tutorial</title>
  </head>
  <body>
    <!-- Include Scala.js compiled code -->
    <script type="text/javascript" src="./target/scala-2.11/scala-js-tutorial-fastopt.js"></script>
    <!-- Run tutorial.webapp.TutorialApp -->
    <script type="text/javascript">
      tutorial.webapp.TutorialApp().main();
    </script>
  </body>
</html>

The first script tag simply includes the generated code (attention, you might need to adapt the Scala version from 2.11 to 2.10 here if you are using Scala 2.10.x instead of 2.11.x).

In the second script tag, we first get the TutorialApp object. Note the (): TutorialApp is a function in JavaScript, since potential object initialization code needs to be run upon the first access to the object. We then simply call the main method on the TutorialApp object.

Since TutorialApp extends JSApp, the object itself and its main method are automatically made available to JavaScript. This is not true in general. Continue reading this tutorial or have a look at the Export Scala.js API to JavaScript guide for details.

If you now open the newly created HTML page in your favorite browser, you will see … nothing. The println in the main method goes right to the JavaScript console, which is not shown by default in a browser. However, if you open the JavaScript console (e.g. in Chrome: right click -> Inspect Element -> Console) you can see the HelloWorld message.

Step 3: Using the DOM

As the last step has shown, running JavaScript inside an HTML page is not particularly useful if you cannot interact with the page. That’s what the DOM API is for.

Adding the DOM Library

To use the DOM, it is best to use the statically typed Scala.js DOM library. To add it to your sbt project, add the following line to your build.sbt:

libraryDependencies += "org.scala-js" %%% "scalajs-dom" % "0.8.0"

sbt-savvy folks will notice the %%% instead of the usual %%. It means we are using a Scala.js library and not a normal Scala library. Have a look at the Dependencies guide for details. Don’t forget to reload the build file if sbt is still running:

> reload
[info] Loading global plugins from (...)
[info] Loading project definition from (...)/scala-js-tutorial/project
[info] Set current project to Scala.js Tutorial (in build (...)/scala-js-tutorial/)

If you are using an IDE plugin, you will also have to regenerate the project files for autocompletion to work.

Using the DOM Library

Now that we added the DOM library, let’s adapt our HelloWorld example to add a <p> tag to the body of the page, rather than printing to the console.

First of all, we import a couple of things:

import org.scalajs.dom
import dom.document

dom is the root of the JavaScript DOM and corresponds to the global scope of Script (aka the window object). We additionally import document (which corresponds to document in JavaScript) for convenience.

We now create a method that allows us to append a <p> tag with a given text to a given node:

def appendPar(targetNode: dom.Node, text: String): Unit = {
  val parNode = document.createElement("p")
  val textNode = document.createTextNode(text)
  parNode.appendChild(textNode)
  targetNode.appendChild(parNode)
}

Replace the call to println with a call to appendPar in the main method:

def main(): Unit = {
  appendPar(document.body, "Hello World")
}

Rebuild the JavaScript

To rebuild the JavaScript, simply invoke fastOptJS again:

> fastOptJS
[info] Compiling 1 Scala source to (...)/scala-js-tutorial/target/scala-2.11/classes...
[info] Fast optimizing (...)/scala-js-tutorial/target/scala-2.11/scala-js-tutorial-fastopt.js
[success] (...)

As you can see from the log, sbt automatically detects that the sources must be recompiled before fast optimizing.

You can now reload the HTML in your browser and you should see a nice “Hello World” message.

Re-typing fastOptJS each time you change your source file is cumbersome. Luckily sbt is able to watch your files and recompile as needed:

> ~fastOptJS
[success] (...)
1. Waiting for source changes... (press enter to interrupt)

From this point in the tutorial we assume you have an sbt with this command running, so we don’t need to bother with rebuilding each time.

Step 4: Reacting on User Input

This step shows how you can add a button and react to events on it by still just using the DOM (we will use jQuery in the next step). We want to add a button that adds another <p> tag to the body when it is clicked.

We start by adding a method to TutorialApp which will be called when the button is clicked:

@JSExport
def addClickedMessage(): Unit = {
  appendPar(document.body, "You clicked the button!")
}

You will notice the @JSExport annotation. It tells the Scala.js compiler to make that method callable from JavaScript. We must also import this annotation:

import scala.scalajs.js.annotation.JSExport

To find out more about how to call Scala.js methods from JavaScript, have a look at the Export Scala.js API to JavaScript guide.

Since we now have a method that is callable from JavaScript, all we have to do is add a button to our HTML and set its onclick attribute (make sure to add the button before the <script> tags):

<button id="click-me-button" type="button" onclick="tutorial.webapp.TutorialApp().addClickedMessage()">Click me!</button>

Reload your HTML page (remember, sbt compiles your code automatically) and try to click the button. It should add a new paragraph saying “You clicked the button!” each time you click it.

Step 5: Using jQuery

Larger web applications have a tendency to set up reactions to events in JavaScript rather than specifying attributes. We will transform our current mini-application to use this paradigm with the help of jQuery. Also we will replace all usages of the DOM API with jQuery.

Depending on jQuery

Just like for the DOM, there is a typed library for jQuery available in Scala.js. Replace the libraryDependencies += .. line in your build.sbt by:

libraryDependencies += "be.doeraene" %%% "scalajs-jquery" % "0.8.0"

Since we won’t be using the DOM directly, we don’t need the old library anymore. Note that the jQuery library internally depends on the DOM, but we don’t have to care about this. sbt takes care of it automatically.

Don’t forget to reload the sbt configuration now:

  1. Hit enter to abort the ~fastOptJS command
  2. Type reload
  3. Start ~fastOptJS again

Again, make sure to update your IDE project files if you are using a plugin.

Using jQuery

In TutorialApp.scala, remove the imports for the DOM, and add the import for jQuery:

import org.scalajs.jquery.jQuery

This allows you to easily access the jQuery object (usually referred to as $ in JavaScript) in your code.

We can now remove appendPar and replace all calls to it by the simple:

jQuery("body").append("<p>[message]</p>")

Where [message] is the string originally passed to appendPar.

If you try to reload your webpage now, it will not work (typically a TypeError would be reported in the console). The problem is that we haven’t included the jQuery library itself, which is a plain JavaScript library.

Adding JavaScript libraries

An option is to include jquery.js from an external source, such as jsDelivr.

<script type="text/javascript" src="http://cdn.jsdelivr.net/jquery/2.1.1/jquery.js"></script>

This can easily become very cumbersome, if you depend on multiple libraries. The Scala.js sbt plugin provides a mechanism for libraries to declare the plain JavaScript libraries they depend on and bundle them in a single file. All you have to do is activate this and then include the file.

In your build.sbt, set:

skip in packageJSDependencies := false

After reloading and rerunning fastOptJS, this will create scala-js-tutorial-jsdeps.js containing all JavaScript libraries next to the main JavaScript file. We can then simply include this file and don’t need to worry about JavaScript libraries anymore:

<!-- Include JavaScript dependencies -->
<script type="text/javascript" src="./target/scala-2.11/scala-js-tutorial-jsdeps.js"></script>

Setup UI in Scala.js

We still want to get rid of the onclick attribute of our <button>. After removing the attribute, we add the setupUI method, in which we use jQuery to add an event handler to the button. We also move the “Hello World” message into this function.

def setupUI(): Unit = {
  jQuery("#click-me-button").click(addClickedMessage _)
  jQuery("body").append("<p>Hello World</p>")
}

Since we do not call addClickedMessage from plain JavaScript anymore, we can remove the @JSExport annotation (and the corresponding import).

Finally, we add a last call to jQuery in the main method, in order to execute setupUI, once the DOM is loaded:

def main(): Unit = {
  jQuery(setupUI _)
}

Again, since we are not calling setupUI directly from plain JavaScript, we do not need to export it (even though jQuery will call it).

We now have an application whose UI is completely setup from within Scala.js. The next step will show how we can test this application.

Step 6: Testing

In this section we will show how such an application can be tested using uTest, a tiny testing framework which compiles to both Scala.js and Scala JVM. As a note aside, this framework is also a good choice to test libraries that cross compile. See our cross compilation guide for details.

Supporting the DOM

Before we start writing tests which we will be able to run through the sbt console, we first have to solve another issue. Remember the task run? If you try to invoke it now, you will see something like this:

> run
[info] Running tutorial.webapp.TutorialApp
org.mozilla.javascript.EcmaError: ReferenceError: "window" is not defined. (/home/ts/.ivy2/cache/org.webjars/jquery/jars/jquery-1.10.2.jar#META-INF/resources/webjars/jquery/1.10.2/jquery.js#14)
(...)
[trace] Stack trace suppressed: run last compile:run for the full output.
[error] (compile:run) Exception while running JS code: ReferenceError: "window" is not defined. (/home/ts/.ivy2/cache/org.webjars/jquery/jars/jquery-1.10.2.jar#META-INF/resources/webjars/jquery/1.10.2/jquery.js#14)
[error] (...)

What basically happens here is that jQuery (yes, it is included automatically) tries to access the window object of the DOM, which doesn’t exist by default in the Rhino and Node.js runners. To make the DOM available, add the following to your build.sbt:

jsDependencies += RuntimeDOM

If you use the fastOpt stage, this will switch to running your code with PhantomJS instead of Node.js, which you need to install. Otherwise, in Rhino, a fake DOM is automatically made available.

After reloading, you can invoke run successfully:

> run
[info] Running tutorial.webapp.TutorialApp
[success] (...)

Just like other library dependencies, this setting applies transitively: if you depend on a library that depends on the DOM, then you depend on the DOM as well.

Adding uTest

Using a testing framework in Scala.js is not much different than on the JVM. It typically boils down to two sbt settings in the build.sbt file. For uTest, these are:

libraryDependencies += "com.lihaoyi" %%% "utest" % "0.3.0" % "test"

testFrameworks += new TestFramework("utest.runner.Framework")

We are now ready to add a first simple test suite (src/test/scala/tutorial/webapp/TutorialTest.scala):

package tutorial.webapp

import utest._

import org.scalajs.jquery.jQuery

object TutorialTest extends TestSuite {

  // Initialize App
  TutorialApp.setupUI()

  def tests = TestSuite {
    'HelloWorld {
      assert(jQuery("p:contains('Hello World')").length == 1)
    }
  }
}

This test uses jQuery to verify that our page contains exactly one <p> element which contains the text “Hello World” after the UI has been set up.

To run this test, simply invoke the test task:

> test
[info] Compiling 1 Scala source to (...)/scalajs-tutorial/target/scala-2.11/test-classes...
[info] 1/2     tutorial.webapp.TutorialTest.HelloWorld		Success
[info] 2/2     tutorial.webapp.TutorialTest		Success
[info] -----------------------------------Results-----------------------------------
[info] tutorial.webapp.TutorialTest		Success
[info]     HelloWorld		Success
[info] Failures:
[info]
[info] Tests: 2
[info] Passed: 2
[info] Failed: 0
[success] (...)

We have successfully created a simple test. Just like run, the test task uses Rhino by default, or Node.js/PhantomJS in fastOpt stage.

A more complex test

We also would like to test the functionality of our button. For this we face another small issue: the button doesn’t exist when testing, since the tests start with an empty DOM tree. To solve this, we create the button in the setupUI method and remove it from the HTML:

jQuery("""<button type="button">Click me!</button>""")
  .click(addClickedMessage _)
  .appendTo(jQuery("body"))

This brings another unexpected advantage: We don’t need to give it an ID anymore but can directly use the jQuery object to install the on-click handler.

We now define the ButtonClick test just below the HelloWorld test:

'ButtonClick {
  def messageCount =
    jQuery("p:contains('You clicked the button!')").length

  val button = jQuery("button:contains('Click me!')")
  assert(button.length == 1)
  assert(messageCount == 0)

  for (c <- 1 to 5) {
    button.click()
    assert(messageCount == c)
  }
}

After defining a helper method that counts the number of messages, we retrieve the button from the DOM and verify we have exactly one button and no messages. In the loop, we simulate a click on the button and then verify that the number of messages has increased.

You can now call the test task again:

> test
[info] Compiling 1 Scala source to (...)/scalajs-tutorial/target/scala-2.11/test-classes...
[info] 1/3     tutorial.webapp.TutorialTest.HelloWorld		Success
[info] 2/3     tutorial.webapp.TutorialTest.ButtonClick		Success
[info] 3/3     tutorial.webapp.TutorialTest		Success
[info] -----------------------------------Results-----------------------------------
[info] tutorial.webapp.TutorialTest		Success
[info]     HelloWorld		Success
[info]     ButtonClick		Success
[info] Failures:
[info]
[info] Tests: 3
[info] Passed: 3
[info] Failed: 0
[success] (...)

This completes the testing part of this tutorial.

Step 7: Optimizing for Production

Here we show a couple of things you might want to do when you promote your application to production.

Full Optimization

Size is critical for JavaScript code on the web. To compress the compiled code even further, the Scala.js sbt plugin uses the advanced optimizations of the Google Closure Compiler. To run full optimizations, simply use the fullOptJS task:

> fullOptJS
[info] Optimizing (...)/scala-js-tutorial/target/scala-2.11/scala-js-tutorial-opt.js
[info] Closure: 0 error(s), 0 warning(s)
[success] (...)

Note that this can take a while on a larger project (tens of seconds), which is why we typically don’t use fullOptJS during development, but fastOptJS instead. If you have used the FastOptStage before, there is a corresponding FullOptStage to run code after full optimization.

Automatically Creating a Launcher

Before creating another HTML file which includes the fully optimized JavaScript, we are going to introduce another feature of the sbt plugin. Since the sbt plugin is able to detect the JSApp object of the application, there is no need to repeat this in the HTML file. If you add the following setting to your build.sbt, sbt will create a scala-js-tutorial-launcher.js file which calls the main method:

persistLauncher in Compile := true

persistLauncher in Test := false

We set persistLauncher to false for testing, since we do not have an application to run. In our HTML page, we can now include this file instead of the manual launcher:

<!-- Run JSApp -->
<script type="text/javascript" src="./target/scala-2.11/scala-js-tutorial-launcher.js"></script>

If we rename our JSApp object, we need not change our HTML at all anymore. Note that the launcher generation only works if you have a single JSApp object. If you happen to have multiple JSApp objects but still would like to generate a launcher, you can set the JSApp object explicitly. Have a look at the Compiling, Running, Linking, Optimizing guide for more details.

Putting it all Together

We can now create our final production HTML file scalajs-tutorial.html which includes the fully optimized code:

<!DOCTYPE html>
<html>
  <head>
    <meta charset="UTF-8">
    <title>The Scala.js Tutorial</title>
  </head>
  <body>
    <!-- Include JavaScript dependencies -->
    <script type="text/javascript" src="./target/scala-2.11/scala-js-tutorial-jsdeps.js"></script>
    <!-- Include Scala.js compiled code -->
    <script type="text/javascript" src="./target/scala-2.11/scala-js-tutorial-opt.js"></script>
    <!-- Run main object -->
    <script type="text/javascript" src="./target/scala-2.11/scala-js-tutorial-launcher.js"></script>
  </body>
</html>

This completes the Scala.js tutorial. Refer to our documentation page for deeper insights into various aspects of Scala.js.