What’s new in the 0.14.0-RC1 technology preview?

Export as a dual of Import

A new export keyword is added to the language that defines aliases for selected members of an object. Consider the following example:

class BitMap
class InkJet
class Printer {
  type PrinterType
  def print(bits: BitMap): Unit = ???
  def status: List[String] = ???
}
class Scanner {
  def scan(): BitMap = ???
  def status: List[String] = ???
}
class Copier {
  private val printUnit = new Printer { type PrinterType = InkJet }
  private val scanUnit = new Scanner
  export scanUnit.scan
  export printUnit.{status => _, _}
  def status: List[String] = printUnit.status ++ scanUnit.status
}

When defined like this, the Copier class defines aliases of the scanner's scan method and all the methods of printUnit except the status method. You can hence call them on the Copier as follows:

val copier = new Copier
copier.print(copier.scan())

The motivation for this change is to promote composition over inheritance. In OOP languages it is easy to define inheritance but the above example would be tricky if you follow the composition route. One would need to implement proxy methods in the Copier to delegate to those of the Scanner and the Printer.

With the export feature, making the Copier behave as the Printer and the Scanner became much more ergonomic. Also, note the fine-grained control over which methods are exposed in cases of the possible method collision, as shown with the status method example.

For more information, please read more in the documentation.

An immutable array type

A new type, scala.IArray[T], is added, which is an immutable version of the Array type. Its implementation deserves a special attention, as it uses the new Dotty features in an elegant way (the below is an abstract from the corresponding commit:

opaque type IArray[T] = Array[T]

object IArray {

  implied arrayOps {
    inline def (arr: IArray[T]) apply[T] (n: Int): T = (arr: Array[T]).apply(n)
    inline def (arr: IArray[T]) length[T] : Int = (arr: Array[T]).length
  }
  def apply[T: ClassTag](xs: T*): IArray[T] = Array(xs: _*)
  /*...*/
}

Essentially, the above defines a wrapper around the ordinary Array which exposes only its apply (to get an element by index) and length methods:

• opaque type IArray[T] defines a type which is known to be an Array, but this information is known only in its companion object IArray. To the rest of the world, this information is not available.
• The implied arrayOps implied instance defines the extension methods that expose the operations available on the IArray type.
• The extension methods, apply and length, delegate to these of the Array type. These methods are inlined which means the performance footprint is the same as that of the original Array methods.
• Because the methods are inline and because the IArray type is opaque, we need the (arr: Array[T]) casts. IArray is known to be equal to Array only inside the IArray object and the inlining will cause the extension methods' bodies to appear outside the IArray object when these methods are called.

Creator Applications

This new language feature is a generalisation of the ability to construct case classes without the new keyword:

class StringBuilder(s: String) {
  def this() = this("")
}
StringBuilder("abc")  // same as new StringBuilder("abc")
StringBuilder()       // same as new StringBuilder()

The motivation for the feature is mainly ergonomic. To make it possible, a new interpretation was added to a function call f(a). Previously, the rules were as follows:

Given a function call f(args),

• if f is a method applicable to args, typecheck f(args) unchanged,
• otherwise, if f has an apply method applicable to args as a member, continue with f.apply(args),
• otherwise, if f is of the form p.m and there is an implicit conversion c applicable to p so that c(p).m is applicable to args, continue with c(p).m(args)

There's now a fourth rule following these rules:

• otherwise, if f is syntactically a stable identifier, and new f where f is interpreted as a type identifier is applicable to args, continue with new f(args).

For more information, please see the documentation.

Other changes

Some of the other changes include:

• infer method renamed to the, the semantics of which is now the same as that of the the method of Shapeless. Namely, the implicits are resolved more precisely – see this gist for an example in Shapeless, and the Dotty documentation for more details.
• The syntax of quoting and splicing was changed. Now the quoting is expressed via '{ ... } and '[...] and splicing – via ${...} and $id. Please see the documentation for more details on these features.

Let us know what you think!

If you have questions or any sort of feedback, feel free to send us a message on our Gitter channel. If you encounter a bug, please open an issue on GitHub.

Contributing

Thank you to all the contributors who made this release possible!

According to git shortlog -sn --no-merges 0.13.0-RC1..0.14.0-RC1 these are:

   214  Martin Odersky
   151  Nicolas Stucki
    71  Liu Fengyun
    53  Guillaume Martres
    26  Olivier Blanvillain
    10  Aleksander Boruch-Gruszecki
     9  Aggelos Biboudis
     6  Miles Sabin
     4  Allan Renucci
     4  Dale Wijnand
     3  Anatolii Kmetiuk
     2  Fengyun Liu
     2  Alex Zolotko
     1  gnp
     1  tim-zh
     1  Dmitry Petrashko
     1  Dotty CI
     1  Jasper Moeys
     1  Jentsch
     1  Jim Van Horn
     1  Lionel Parreaux
     1  Master-Killer
     1  Olivier ROLAND
     1  Robert Stoll
     1  Seth Tisue
     1  Tomasz Godzik
     1  Victor

If you want to get your hands dirty and contribute to Dotty, now is a good time to get involved! Head to our Getting Started page for new contributors, and have a look at some of the good first issues. They make perfect entry points into hacking on the compiler.

We are looking forward to having you join the team of contributors.

Library authors: Join our community build

Dotty now has a set of widely-used community libraries that are built against every nightly Dotty snapshot. Currently this includes ScalaPB, algebra, scalatest, scopt and squants. Join our community build to make sure that our regression suite includes your library.