A reactive web framework. It enables you use native XML literal syntax to create reactive DOM nodes, which are able to automatically change whenever the data source changes.

Binding.scala

Binding.scala is a data-binding framework for Scala, running on both JVM and Scala.js.

Binding.scala can be used as a reactive templating language in both web and desktop GUI development. It enables you use native XHTML and FXML literal syntax to create reactive DOM nodes and JavaFX components, which are able to automatically change whenever the data source changes.

See Binding.scala • TodoMVC or ScalaFiddle DEMOs as examples for common tasks when working with Binding.scala.

Comparison to other reactive web frameworks

Binding.scala has more features and less concepts than other reactive web frameworks like ReactJS.

Binding.scala ReactJS Support HTML literal? Yes Partially supported. Regular HTML does not compile, unless developers manually replaces class and for attributes to className and htmlFor, and manually converts inline styles from CSS syntax to JSON syntax. Algorithm to update DOM Precise data-binding, which is faster than virtual DOM Virtual DOM differentiation, which requires manually managed key attributes for complicated DOM. Lifecycle management for data-binding expressions Automatically N/A Statically type checking Yes, even for HTML tags and attribues No Learning curve Always easy Easy to start. Requires much more efforts to understand its corner cases.

See Design section for more information.

Getting started

We will build an Binding.scala web page during the following steps.

Step 0: Setup a Sbt Scala.js project

See http://www.scala-js.org/tutorial/basic/ for information about how to setup such a project.

Step 1: Add Binding.scala dependencies into your build.sbt:

libraryDependencies += "com.thoughtworks.binding" %%% "dom" % "latest.release"

addCompilerPlugin("org.scalamacros" % "paradise" % "2.1.0" cross CrossVersion.full)

Step 2: Create a data field, which contains some Var and Vars as data source for your data-binding expressions

case class Contact(name: Var[String], email: Var[String])

val data = Vars.empty[Contact]

A Var represents a bindable variable, which also implements Binding trait, hence a Var can be seen as a binding expression as well. If another expression depends on a Var, the value of the expression changes whenever value of the Var changes.

A Vars represents a sequence of bindable variables, which also implements BindingSeq trait, hence a Vars can be seen as a binding expression of a sequence as well. If another comprehension expression depends on a Vars, the value of the expression changes whenever value of the Vars changes.

Step 3: Create a @dom method that contains data-binding expressions

@dom
def table: Binding[Table] = {
  <table border="1" cellPadding="5">
    <thead>
      <tr>
        <th>Name</th>
        <th>E-mail</th>
      </tr>
    </thead>
    <tbody>
      {
        for (contact <- data) yield {
          <tr>
            <td>
              {contact.name.bind}
            </td>
            <td>
              {contact.email.bind}
            </td>
          </tr>
        }
      }
    </tbody>
  </table>
}

A @dom method represents a data-binding expression.

The return type is always wrapped in a com.thoughtworks.binding.Binding trait. For example @dom def x: Binding[Int] = 1, @dom def message: Binding[String] = "content"

A @dom method supports HTML literal. Unlike normal XML literal in a normal Scala method, the types of HTML literal are specific subtypes of org.scalajs.dom.raw.Node or com.thoughtworks.binding.BindingSeq[org.scalajs.dom.raw.Node], instead of scala.xml.Node or scala.xml.NodeSeq. So we could have @dom def node: Binding[org.scalajs.dom.raw.HTMLBRElement] = <br/> and @dom def node: Binding[BindingSeq[org.scalajs.dom.raw.HTMLBRElement]] = <br/><br/>.

A @dom method is composed with other data-binding expressions in two ways:

1. You could use bind method in a @dom method to get value of another Binding.
2. You could use for / yield expression in a @dom method to map a BindingSeq to another.

You can nest Node or BindingSeq[Node] in other HTML element literals via { ... } interpolation syntax.

Step 4: Render the data-binding expressions to DOM in the main method

@JSExport
def main(): Unit = {
  dom.render(document.body, table)
}

Step 5: Invoke the main method in a HTML page

<!DOCTYPE html>
<html>
  <head>
    <script type="text/javascript" src="js-fastopt.js"></script>
  </head>
  <body>
    <script type="text/javascript">
      SampleMain().main()
    </script>
  </body>
</html>

Now you will see a table that just contains a header, because data is empty at the moment.

Step 6: Add some <button> to fill data for the table

@dom
def table: Binding[BindingSeq[Node]] = {
  <div>
    <button
      onclick={ event: Event =>
        data.value += Contact(Var("Yang Bo"), Var("yang.bo@rea-group.com"))
      }
    >
      Add a contact
    </button>
  </div>
  <table border="1" cellPadding="5">
    <thead>
      <tr>
        <th>Name</th>
        <th>E-mail</th>
        <th>Operation</th>
      </tr>
    </thead>
    <tbody>
      {
        for (contact <- data) yield {
          <tr>
            <td>
              {contact.name.bind}
            </td>
            <td>
              {contact.email.bind}
            </td>
            <td>
              <button
                onclick={ event: Event =>
                  contact.name.value = "Modified Name"
                }
              >
                Modify the name
              </button>
            </td>
          </tr>
        }
      }
    </tbody>
  </table>
}

When you click the "Add a contact" button, it appends a new Contact into data, then, Binding.scala knows the relationship between DOM and data, so it decides to append a new <tr> corresponding to the newly appended Contact.

And when you click the "Modify the name", the name field on contact changes, then, Binding.scala decides to change the content of the corresponding tr to new value of name field.

See https://github.com/ThoughtWorksInc/Binding.scala-sample for the complete example.

Design

Precise data-binding

ReactJS requires users to provide a render function for each component. The render function should map props and state to a ReactJS's virtual DOM, then ReactJS framework creates a DOM with the same structure as the virtual DOM.

When state changes, ReactJS framework invokes render function to get a new virtual DOM. Unfortunately, ReactJS framework does not precisely know what the state changing is. ReactJS framework has to compare the new virtual DOM and the original virtual DOM, and guess the changeset between the two virtual DOM, then apply the guessed changeset to the real DOM as well.

For example, after you prepend a table row <tr> into an existing <tbody> in a <table>, ReactJS may think you also changed the content of every existing <tr> of the <tbody>.

The reason for this is that the render function for ReactJS does not describe the relationship between state and DOM. Instead, it describes the process to create a virtual DOM. As a result, the render function does not provide any information about the purpose of the state changing, although a data-binding framework should need the information.

Unlike ReactJS, a Binding.scala @dom method is NOT a regular function. It is a template that describes the relationship between data source and the DOM. When part of the data source changes, Binding.scala knows about the exact corresponding partial DOM affected by the change, thus only re-evaluating that part of the @dom method to reflect the change in the DOM.

With the help of the ability of precise data-binding provided by Binding.scala, you can get rid of concepts for hinting ReactJS's guessing algorithm, like key attribute, shouldComponentUpdate method, componentDidUpdate method or componentWillUpdate method.

Composability

The smallest composable unit in ReactJS is a component. It is fair to say that a React component is lighter than an AngularJS controller, while Binding.scala is better than that.

The smallest composable unit in Binding.scala is a @dom method. Every @dom method is able to compose other @dom methods via .bind.

case class Contact(name: Var[String], email: Var[String])

@dom
def bindingButton(contact: Contact): Binding[Button] = {
  <button
    onclick={ event: Event =>
      contact.name.value = "Modified Name"
    }
  >
   Modify the name
  </button>
}

@dom
def bindingTr(contact: Contact): Binding[TableRow] = {
  <tr>
    <td>{ contact.name.bind }</td>
    <td>{ contact.email.bind }</td>
    <td>{ bindingButton(contact).bind }</td>
  </tr>
}

@dom
def bindingTable(contacts: BindingSeq[Contact]): Binding[Table] = {
  <table>
    <tbody>
      {
        for (contact <- contacts) yield {
          bindingTr(contact).bind
        }
      }
    </tbody>
  </table>
}

@JSExport
def main(): Unit = {
  val data = Vars(Contact(Var("Yang Bo"), Var("yang.bo@rea-group.com")))
  dom.render(document.body, bindingTable(data))
}

You may find out this approach is much simpler than ReactJS, as:

• Instead of passing props in ReactJS, you just simply provide parameters for Binding.scala.
• Instead of specifying propTypes in ReactJS, you just simply define the types of parameters in Binding.scala.
• Instead of raising a run-time error when types of props do not match in ReactJS, you just check the types at compile-time.

Lifecycle management for data-binding expressions

The ability of precise data-binding in Binding.scala requires listener registrations on the data source. Other reactive frameworks that have the ability ask users manage the lifecycle of data-binding expressions.

For example, MetaRx provides a dispose method to unregister the listeners created when building data-binding expressions. The users of MetaRx have the responsibility to call dispose method for every map and flatMap call after the expression changes, otherwise MetaRx leaks memory. Unfortunately, manually disposeing everything is too hard to be right for complicated binding expressions.

Another reactive web framework Widok did not provide any mechanism to manage lifecycle of of data-binding expressions. As a result, it simply always leaks memory.

In Binding.scala, unlike MetaRx or Widok, all data-binding expressions are pure functional, with no side-effects. Binding.scala does not register any listeners when users create individual expressions, thus users do not need to manually unregister listeners for a single expression like MetaRx.

Instead, Binding.scala creates all internal listeners together, when the user calls dom.render or Binding.watch on the root expression. Note that dom.render or Binding.watch manages listeners on all upstream expressions, not only the direct listeners of the root expression.

In brief, Binding.scala separates functionality in two kinds:

• User-defined @dom methods, which produce pure functional expressions with no side-effects.
• Calls to dom.render or Binding.watch, which manage all side-effects automatically.

HTML literal and statically type checking

As you see, you can embed HTML literals in @dom methods in Scala source files. You can also embed Scala expressions in braces in content or attribute values of the HTML literal.

@dom
def notificationBox(message: String): Binding[Div] = {
  <div class="notification" title={ s"Tooltip: $message" }>
    {
      message
    }
  </div>
}

Despite the similar syntax of HTML literal between Binding.scala and ReactJS, Binding.scala creates real DOM instead of ReactJS's virtual DOM.

In the above example, <div>...</div> creates a DOM element with the type of org.scalajs.dom.html.Div. Then, the magic @dom lets the method wrap the result as a Binding.

You can even assign the Div to a local variable and invoke native DOM methods on the variable:

@dom
def notificationBox(message: String): Binding[Div] = {
  val result: Div = <div class="notification" title={ s"Tooltip: $message" }>
    {
      message
    }
  </div>

  result.scrollIntoView()

  result
}

scrollIntoView method will be invoked when the Div is created. If you invoke another method not defined in Div, the Scala compiler will report a compile-time error instead of bringing the failure to run-time, because Scala is a statically typed language and the Scala compiler understands the type of Div.

You may also notice class and title. They are DOM properties or HTML attributes on Div. They are type-checked by Scala compiler as well.

For example, given the following typo method:

@dom
def typo = {
  val myDiv = <div typoProperty="xx">content</div>
  myDiv.typoMethod()
  myDiv
}

The Scala compiler will report errors like this:

typo.scala:23: value typoProperty is not a member of org.scalajs.dom.html.Div
        val myDiv = <div typoProperty="xx">content</div>
                     ^
typo.scala:24: value typoMethod is not a member of org.scalajs.dom.html.Div
        myDiv.typoMethod()
              ^

With the help of the static type system, @dom methods can be much more robust than ReactJS components.

You can find a complete list of supported properties and methods on scaladoc of scalajs-dom or MDN

Custom attributes

If you want to suppress the static type checking of attributes, add a data: prefix to the attribute:

@dom def myCustomDiv = {
  val myDiv = <div data:customAttributeName="attributeValue"></div>
  assert(myDiv.getAttribute("customAttributeName") == "attributeValue")
  myDiv
}

The Scala compiler will not report errors now.

Showcases

• TodoMVC: a project which offers the same Todo application implemented using MV* concepts in most of the popular JavaScript MV* frameworks of today.
• Granblue Raid Finder: a site for finding Granblue Fantasy raid tweets.
• Game of Life: Conway's Game of Life implemented with Binding.scala.
• playground-binding.scala : Various DEMOs with scala, scalajs and binding.scala
• CITE Application: A single-page browser application for exploring citable resources.
• hmt-reader: A package of application and data for reading Homer Multitext textual data, in its current release.
• Full-Stack-Scala-Starter: Play 2.5, ScalaJS, Binding.scala starter project.
• scala-adapters: A simple framework to implement your server jobs - providing a standard UI-client to monitor and test them. (Used in Production)
• Binding.scala-Google-Maps: A step-by-step tutorial to get you started with Binding.scala.
• scala-adapters.g8: A Giter8 template for a full-stack Scala project that uses scala-adapters.
• play-akka-telegrambot4s-incidents: An incident management app - where you can send incidents with a chat bot.
  
• scala-adapters-images: A demo project that uses scala-adapters and play-akka-telegrambot4s.
• jmh-view: An embeddable JMH report viewer, which is used to to create the report for the benchmarks of Scala parsers.
• RL-Playground: A web-based interactive reinforcement learning demonstration for games.
• Word Cloud Generator: A browser extension to generate word cloud visualizations of web pages, text files, or other arbitary text inputs.

(Feel free to add your project here)

Downloads

Binding.scala has an extremely tiny code base. The source files are split into few libraries, one file per library.

Core data-binding expressions (Binding.scala)

This module is available for both JVM and Scala.js. You could add it in your build.sbt.

// For JVM projects
libraryDependencies += "com.thoughtworks.binding" %% "binding" % "latest.release"

addCompilerPlugin("org.scalamacros" % "paradise" % "2.1.0" cross CrossVersion.full)

// For Scala.js projects, or JS/JVM cross projects
libraryDependencies += "com.thoughtworks.binding" %%% "binding" % "latest.release"

addCompilerPlugin("org.scalamacros" % "paradise" % "2.1.0" cross CrossVersion.full)

HTML DOM integration (dom.scala)

This module is only available for Scala.js. You could add it in your build.sbt.

// For Scala.js projects
libraryDependencies += "com.thoughtworks.binding" %%% "dom" % "latest.release"

addCompilerPlugin("org.scalamacros" % "paradise" % "2.1.0" cross CrossVersion.full)

FXML integration (fxml.scala)

This module is available for both JVM and Scala.js. You could add it in your build.sbt.

// For JVM projects
libraryDependencies += "com.thoughtworks.binding" %% "fxml" % "latest.release"

addCompilerPlugin("org.scalamacros" % "paradise" % "2.1.0" cross CrossVersion.full)

// For Scala.js projects, or JS/JVM cross projects
libraryDependencies += "com.thoughtworks.binding" %%% "fxml" % "latest.release"

addCompilerPlugin("org.scalamacros" % "paradise" % "2.1.0" cross CrossVersion.full)

Remote data-binding for scala.concurrent.Future (FutureBinding.scala)

This module is available for both JVM and Scala.js. You could add it in your build.sbt.

// For JVM projects
libraryDependencies += "com.thoughtworks.binding" %% "futurebinding" % "latest.release"

addCompilerPlugin("org.scalamacros" % "paradise" % "2.1.0" cross CrossVersion.full)

// For Scala.js projects, or JS/JVM cross projects
libraryDependencies += "com.thoughtworks.binding" %%% "futurebinding" % "latest.release"

addCompilerPlugin("org.scalamacros" % "paradise" % "2.1.0" cross CrossVersion.full)

Remote data-binding for ECMAScript 2015's Promise (JsPromiseBinding.scala)

This module is only available for Scala.js. You could add it in your build.sbt.

// For Scala.js projects
libraryDependencies += "com.thoughtworks.binding" %%% "jspromisebinding" % "latest.release"

addCompilerPlugin("org.scalamacros" % "paradise" % "2.1.0" cross CrossVersion.full)

Links

• The API documentation
• Binding.scala • TodoMVC
• ScalaFiddle DEMOs
• Binding.scala Questions on Stack Overflow
• 本README的中文版