Scala for Java Devs

Scala for Java Devs
Christos KK Loverdos
loverdos at github, twitter, gmail
Java meetup @ Thessaloniki
2017‐10‐12
Java meetup @ Thessaloniki - 2017-10-12

About me
Software engineer, nearly 20 years now
Architect, team lead, technical PM
Telcos, startups
employee, freelancer, consultant
Java enthusiast since 1997
Scala enthusiast since 2004 (2007)
Co-author of "Steps in Scala"
Java meetup @ Thessaloniki - 2017-10-12 2

A very incomplete meline
2004 v1.0
2006 v2.0
2007 v2.7 (+Li web framework)
2010 v2.8 (+ScalaDays @ EPFL, 180 participants)
2011 v2.9
2012 v2.10
now v2.12 (+Java 8 interoperability)

Main ideas ‐ Why Scala?
Synthesis of OOP and FP
Being scalable
Programming in the small vs programming in
the large
Provide the right abstractions in the core,
everything else in libraries
Rich type system

A few highlights
Immutable objects and collections
Type inference
Func ons are rst-class
Everything is an object
Pattern matching
Domain Speci c Languages
REPL

Warm‐up

Hello world
hello.scala
object hello {
def main(args: Array[String]): Unit =
println("Hello world")
}

Hello world
hello.scala
object hello {
}
$ scala -save hello.scala
Hello world

Hello world
hello.scala
object hello {
}
$ scala -save hello.scala
Hello world
$ ls hello.*
hello.jar hello.scala

Hello world Scala vs Java
object hello {}
public class hello {}
object, class
No public

def main(...): Unit = {}
public static void main(...) {}
Unit vs void
No sta c
Return type a er vs before method name & args
Scala object implies Java sta c

args: Array[String]
String[] args
Type a er vs before the name

args: Array[String]
String[] args
Type a er vs before the name
DIM X AS INTEGER
VAR X: Integer

REPL: a value is an object

A simple class
class Complex(val re: Double, val im: Double) {
def this() = this(0, 0)
def mag = Math.sqrt(re*re + im*im)
def polarCoordinates = {
val r = Math.sqrt(re*re + im*im)
val phi = Math.atan2(im, re)
(r, phi)
}
override def toString = s"Complex($re, $im)"
}

A simple class
class Complex(val re: Double, val im: Double) {
def this() = this(0, 0)
def mag: Double = Math.sqrt(re*re + im*im)
def polarCoordinates: (Double, Double) = {
val r = Math.sqrt(re*re + im*im)
val phi = Math.atan2(im, re)
(r, phi)
}
override def toString: String = s"Complex($re, $im)"
}

Super constructor
class Animal(name: String)
class Dog(name: String) extends Animal(name)

Some more fun

Trait (interface)
trait Printer {
def printPDF(pdf: File): Unit
def printRTF(rtf: File): Unit = {
val pdf = ... // convert RTF to PDF
printPDF(pdf)
}
}
Using default methods
(https://github.com/scala/scala/pull/5003)
as of Scala 2.12

Trait (mixin)
trait Mage {
def castSpell(spell: Spell, target: Target) = ...
}
trait Fighter {
def useSword(target: Target) = ...
}
class Player1 extends Mage
class Player2 extends Mage with Fighter

Case class
case class Complex(re: Double, im: Double)

Case class
val c = new Complex(1.2, 2.0)
val cc = Complex(1.2, 2.0)
No need to use new

Case class
val c = new Complex(1.2, 2.0)
val cc = Complex(1.2, 2.0)
// (c == cc) is true
Automatic, derived, structural equality
Compiler implements hashCode and equals
Very handy for immutable domain objects

Case class
// instead of
// case class Complex(val re: Double, val im: Double)
Constructor arguments are promoted to class
attributes

Case class
val zero = Complex(0, 0)
val one = zero.copy(re = 1)
val i = zero.copy(im = 1)
Builtin copy()

Case class ‐ Pa ern matching
val c: Complex = ...
c match {
case Complex(0, 0) => // zero
case Complex(1, 0) => // real unit
case Complex(0, 1) => // imaginary unit
case Complex(a, b) if a == b => // equal coordinates
case Complex(a, b) => // all other cases ...
}

Case class ‐ ASTs
ast.scala
sealed trait AstNode
case class Expr(n: Number) extends AstNode
case class Add(left: Expr, right: Expr) extends AstNode
other.scala
val x: AstNode = parseSourceCode(...)
x match {
case Expr(n) => ...
case Add(left, right) => ...
}

Generics
type parameter in a class
trait Ordered[T] {
def < (that: T): Boolean
def <=(that: T): Boolean
...
}
class ANumber(x: Int) extends Ordered[ANumber] { ... }
Ordered[T] characterizes a type T that has a single,
natural ordering.

Generics
Collections
final class Array[T] // mutable
sealed abstract class List[T] // immutable
val a = Array(1, 2, 3)
val b = List(1, 2, 3)
val c = 1 :: 2 :: 3
Java's Collection<T> becomes Collection[T]

digression: List again

digression: List again
sealed abstract class List[+A] extends AbstractSeq[A]
with LinearSeq[A]
with Product
with GenericTraversableTemplate[A, List]
with LinearSeqOptimized[A, List[A]]
with Serializable

Generics (method)
Parametric polymorphism
object Sorter {
def quickSort[T](elems: Array[T], /*what else?*/)
}

Now that we've talked about
generics, let's talk about
func ons

Func on (deﬁni on)
trait Function1[-A, +B] extends AnyRef {
def apply(a: A): B
}

Func on (deﬁni on)
trait Function1[A, B] {
def apply(a: A): B
}
We can also represent the type Function1[A, B] as
A => B or
(A) => B
This is a total function

Func on
Math:
f : A → B
Scala:
f: A => B
f: Function1[A, B]
Java:
Function1<A, B> f

Func on (declara on)
val s_length:
(String) => Int =
(s: String) => s.length
Notice how the method length of class String was
promoted to a function.

Func on (declara on)
val s_length: (String) => Int = s => s.length
// or
val s_length: (String) => Int = _.length
With some nice syntactic sugar

Func on (applica on)
We apply a function f of type A => B
val f: A => B = ...
to a value of type A
val x: A = ...
using intuitive syntax
f(x)

... which gets desugared to
f.apply(x)
Remember that Function1 is de ned with one
method:
trait Function1[A, B] {
def apply(a: A): B
}

... which gets desugared to
f.apply(x)
Everything is an object

apply is a binary method and you can also write
f apply x
Same thing for computing e.g. the maximum of two
integers
x max y vs x.max(y)

Collec ons + Func ons = fun

map
Map each element of a collection to a new
element, according to a well de ned function
trait Collection[A] {
def map[B](f: A => B): Collection[B]
}

map
}
scala> List("a", "bb").map(x => x.length)
res2: List[Int] = List(1, 2)

map
}
scala> List(1, 2, 3).map( n => isOdd(n) )
res3: List[Boolean] = List(true, false, true)

map
}
alist.map(x => isOdd(x))
alist.map( isOdd(_) )
alist.map( isOdd )
alist map isOdd

... from map to Map
The generic type is Map[A, B]
The default implementation is immutable
val numbers: Map[Int, String] = Map()
or
val numbers = Map[Int, String]()

... from map to Map
Initialization is not verbose
val numbers = Map[Int, String](
1 -> "one",
2 -> "two",
3 -> "three"
)

... and then to groupBy
Now that we have Lists and Maps
scalal> val list = List(1, 2, 3)
list: List[Int] = List(1, 2, 3)
scala> val isOdd = (x: Int) => x % 2 == 1
isOdd: Int => Boolean = <function>
scala> val oddeven = list.groupBy(isOdd)
oddeven: scala.collection.immutable.Map[Boolean,List[Int]] =
Map(false -> List(2), true -> List(1, 3))

What is the type of groupBy?

class List[A] {
def groupBy ...
}

class List[A] {
def groupBy[B](f: A => B): ???
}

class List[A] {
def groupBy[B](f: A => B): Map[B, List[A]]
}

null = Billion dollar mistake

Op on
trait Option[+T]
case class Some[T](t: T) extends Option[T]
case object None extends Option[Nothing]
Nothing is a subtype of every other type but it has no
instances

Op on
scala> val xOpt: Option[Complex] = Option(null)
xOpt: Option[Complex] = None
scala> val x = xOpt.getOrElse(Complex(0, 0))
x: Complex = Complex(0.0,0.0)

Op on (pa ern match)
xOpt match {
case Some(Complex(re, im)) => ...
case None =>
}

Op on (for comprehension)
for {
x <- xOpt
} { ... }
If we only care about the Some() case.
In effect, you can view Option as a simple
collection of at most one item.

For comprehension
for {
item <- List(1, 2, 3)
} yield item * 2
Computes a new list with each item doubled.
This is a map in disguise.

For comprehension
for {
item <- List(1, 2, 3)
} yield item * 2
same as
List(1, 2, 3).map(_ * 2)

For comprehension
for {
item <- List(1, 2, 3)
} yield item * 2
same as
List(1, 2, 3).map(_ * 2)
Remember that
List(1, 2, 3).map(x => x * 2)

Add some more ingredients,
and then you get

?

Monads! (in another talk)
for {
x <- Some(Complex(0.0, 1.0))
m <- List(1.0, 2.0, 3.0)
} yield Complex(x.re * m, x.im * m)

Monads! (in another talk)
for {
x <- Some(Complex(0.0, 1.0))
m <- List(1.0, 2.0, 3.0)
What about this?
for {
x <- None
m <- List(1.0, 2.0, 3.0)

And since we are talking
about op onal things ...

Op onal arguments (class)
case class Collector(
name: String,
coins: List[Coin] = List()
books: List[Book] = List() // = Nil
)

name: String,
)
val collector1 = Collector(
"John Smith"
)

name: String,
)
"John Smith",
List(OneEuroCoin),
List(Book("Lord of the rings"))
)

name: String,
)
name = "John Smith",
coins = List(OneEuroCoin),
books = List(Book("Lord of the rings"))
)

Not covered

Easy stuﬀ
Try[T] data type
Future[T] data type
lazy values
by-name parameters
...

Not so easy stuﬀ
Variance (covariance, contravariance)
List[+T]
Type constructors, higher kinds
(very informally)
List : T → List[T]
Implicits
real power (cf. Haskell type classes).

Epilogue

Opinions circa 2009
"If I were to pick a language to use today other
than Java, it would be Scala"
James Gosling, creator of Java
"If someone had shown me the 'Programming in
Scala' book back in 2003, I'd probably have never
created Groovy"
James Strachan, creator of Groovy

Resources
scala-lang.org
google.com/search?q=awesome-scala
github.com/trending/scala

Thank you!

Slides that did not survive

BEGIN warm‐up

END warm‐up

BEGIN war map

END war map

Where to ﬁnd me
select username
from venue
where username = 'loverdos'
and venuename in ('gmail', 'twitter', 'github')

Scala for Java Devs

More Related Content

What's hot (20)

Similar to Scala for Java Devs (20)

Recently uploaded (20)

Scala for Java Devs