Introduction to Scala for Java Programmers

An Introduction to
Scala for Java
Programmers
Adam Rabung

Thursday, November 25, 2010

In 2003, Martin Odersky and his team set
out to create a “scalable” language that
fused important concepts from functional
and object-oriented programming. The
language aimed to be highly interoperable
with Java, have a rich type system, and to
allow developers to express powerful
concepts concisely.


Scalable Language


Scalable Language
object Hello {
def main(args:Array[String]) {
println("Hello World")
}
}


Scalable Language

mention linkedin connect with jruby front and scala back

mention using clojure data structures in scala


Scala loves Java
import javax.servlet.http.HttpServlet
import javax.servlet.http.HttpServletRequest
import javax.servlet.http.HttpServletResponse

class ScalaServlet extends HttpServlet {

override def doGet(request: HttpServletRequest,
response: HttpServletResponse) {

response.getWriter().println("Hello World!")
}

}


Java loves Scala
class CrazyScalaTechnology {
def doWebThreeDotOhStuff(msg:String) {
println(msg) Scala
}
}

public class JavaPojo { Java
public static void main(String[] args) {
new CrazyScalaTechnology().doWebThreeDotOhStuff("Hello world!");
}
}

eclipse supports mixed projects that can hold .scala and .java

note the String being passed in to Java - in fact, most of the scala “stdlib” is just
javas


Highly interoperable
with Java
• To your boss and the people who
administer your servers, it’s “just a jar”
• Because it’s not an all or nothing
proposition, you can slowly “sneak” Scala
into your codebase - think test cases!


Scala is OO

• Classes/Interfaces/Fields
• Everything is an object
• Traits
• Singletons


Basics
class Account(var currentBalance:MonetaryAmount,
val accountOwner:Person,
val withdrawalLimit:MonetaryAmount) {

private val txLog = new TransactionLog()
txLog.log(“New acct with “ + currentBalance)

def getSummary = "Name = " + accountOwner.name +
", Balance = " + currentBalance

def withdraw(amount:MonetaryAmount) {
if (amount < currentBalance && amount < withdrawalLimit)
currentBalance = currentBalance - amount
else
txLog.log("Withdrawal rejected, amount too high: " + amount)
}
}


Basics class name/extensions
field declarations
constructor + copying to
fields

val withdrawalLimit:MonetaryAmount)
extends IAccount {



else
}
}


Basics
extends IAccount {

• var means “variable” - readable and
writeable. Automatically generates getters
and setters
• val means “value” - read-only. Generates
only getters and the ﬁeld is not re-
assignable val does not mean
“immutable”


Basics extra class initialization

private field

extends IAccount {



else
}
}


Basics
val withdrawalLimit:MonetaryAmount) {

body of the class is the
body of the constructor


else
}
}


Basics: Accessing your
ﬁelds

val account = new Account(balance, user, limit)
account.currentBalance = new MonetaryAmount(50)
println("New balance is " + account.currentBalance)

account.accountOwner = new Person("Hacker Johnson")

“Setter” “Getter”
Won’t compile - accountOwner is a “val”


public class Account implements IAccount {
private MonetaryAmount currentBalance;
private final Person accountOwner;
private final MonetaryAmount withdrawalLimit;
private final TransactionLog txLog;

public Account(MonetaryAmount currentBalance,
Person accountOwner,
MonetaryAmount withdrawalLimit
) {
this.currentBalance = currentBalance;
this.accountOwner = accountOwner;
this.withdrawalLimit = withdrawalLimit;
this.txLog = new TransactionLog();
txLog.log("New acct created");
}

public String getSummary() {
return "Name = " + accountOwner.name +
", Balance = " + currentBalance;
}

public void withdraw(MonetaryAmount amount) {
if (amount.isLessThan(currentBalance) &&
amount.isLessThan(withdrawalLimit)) {
currentBalance = currentBalance.minus(amount);
}
else {
txLog.log("Withdrawal rejected, amount too high: "
+ amount);
}
}

public MonetaryAmount getCurrentBalance() {
return currentBalance;
}

public void setCurrentBalance(MonetaryAmount currentBalance) {
this.currentBalance = currentBalance;
}

public Person getAccountOwner() {
return accountOwner;
}

public MonetaryAmount getWithdrawalLimit() {
return withdrawalLimit;
}
}


Everything is an object

• No operators
• No primitives
• No arrays
• No statics did you notice me using <
and - on MonetaryAmt?


OO:
Inheritance
trait Fruit {
def getColor : String
Fruit is an ifc
def hasStem : Boolean
Brief syntax
} single expression methods can skip {

override keyword
class Apple extends Fruit {
def getColor = "Red" default is public

def hasStem = true
}

class GrannySmithApple extends Apple {
override def getColor = "Green"
}


OO: traits

• “Interfaces with implementation”
• Like interfaces, a class can extend many
traits
• Like abstract classes, a trait can deﬁne
abstract and concrete methods - multiple
inheritance!!


OO: traits
Special considerations
• No initialization of ﬁelds with constructors
• Special meaning of “super”:
• In Java, known when you compile the
class
• With traits, known only when you mix it
in


OO: traits
• 3 interesting uses of trait
• Rich Interfaces
• Stackable modiﬁcations/”mixins”
• Compose diverse functionality into a
single unit


Traits: Rich Interfaces
trait Comparable[A] {
def compare(that: A): Int Abstract
def < (that: A): Boolean = (this compare that) < 0
def > (that: A): Boolean = (this compare that) > 0
def <= (that: A): Boolean = (this compare that) <= 0
def >= (that: A): Boolean = (this compare that) >= 0
}

did you notice?
method names w/ special chars
missing dots and parens?

this is just an abstract class, but you dont have to feel guilty about mixing it in


Traits: mixins
/**
* Throw an exception if someone calls get with
* an undefined key
*/
trait NoNullAccessMap[K, V] extends java.util.Map[K, V] {
abstract override def get(key: Object): V = {
val value = super.get(key)
assert(value != null, "No value found for key " + key)
return value
}
}


Traits: mixins
trait NoNullAccessMap[K, V] extends java.util.Map[K, V] {
abstract override def get(key: Object): V = {
val value = super.get(key)
assert(value != null, "No value found for key " + key)
return value
}
}

The exact functionality you are overriding is
not yet known


Traits: mixins
/**
* Throw an exception if someone tries to call put()
* with a key that is already in the map
*/
trait NoOverwriteMap[K, V] extends java.util.Map[K, V] {
abstract override def put(key: K, value: V): V = {
assert(!containsKey(key),
"Could not set " + key + " to " + value +
": it is already set to " + get(key))
return super.put(key, value)
}
}


Traits: mixins (static)

class StrictMap[K, V] extends HashMap[K, V]
with NoNullAccessMap[K, V]
with NoOverwriteMap[K, V]

now that we’re creating a real class, we use HashMap not Map


Traits: mixins (dynamic)

val stateForCity = new HashMap[String, String]()
with NoNullAccessMap[String, String]
with NoOverwriteMap[String, String]

pick and choose
characteristics for
individual map instances


Traits: compose
functionality

class GetAccountsController
extends Controllor
with PermissionCheck
with Logging

of course, java style
composition is fine too


Food for thought

• If you had traits, when would you choose
inheritance and when would you choose
composition?


OO:
singletons
object CheckSums {
def calculateMD5(s:String) = "todo!"
}

def main(args:Array[String]) {
CheckSums.calculateMD5("password")
}

concise way to define a
class and “singletonize”
Item 1:Consider it
update this to use an
interface
providing static
factory methods
instead of
contructors


OO:
singletons
• Concise way to deﬁne a class AND create a
singleton instance of it
• Used for “static” methods
• Thread-safe instantiation
• Extend traits or classes!
• Used exactly like any object

Scala is OO

• Scala is fairly consistent with Java in terms
of simple OO concepts like classes, ﬁelds,
and polymorphism
• Scala takes OO further with concepts like
traits, everything-is-an-object, uniform
access principle, etc.


Scala is functional

• Functions are ﬁrst-class citizens
• Immutability/No Side Effects These concepts are very
important in java of
course

• Pattern matching
• Type inference


Functional
programming

• Imperative programs consist of statements
that change state
• Functional programs evaluate expressions
and avoid using state and mutable data

I no longer make software. I make lenses for the projection of data. :)
@nuttycom


Functional Scala:
functions are objects

def square(x:Int) = x * x


Functional Scala:

val squareFunction = new Function1[Int, Int]() {
def apply(x: Int) = x * x
}

functions can be values
just like Strings and ints

of course, no different
than java here


Functional Scala:
val squareFunction = { x:Int => x * x }

Is just shorthand for
def apply(x:Int) = x * x
}


Functional Scala:
val squareFunction = { x:Int => x * x }

def apply(x:Int) = x * x
}


Imperative Java
public List<String> getShortCityNames(List<String cities) {
List<String> shortCityNames = new ArrayList<String>();
for (String city : cities) {
if (city.length() < 8) {
shortCityNames.add(city);
}
}
return shortCityNames;
}

“imperative” as the opposite of “functional”

you might obser ve whats going on here is we’re taking a list of items, and filtering out those that do not
match a condition


Functional Scala

val cities = List("Richmond", "Charlottesville", "Ashland")
val shortCities = cities.filter { city: String => city.length < 8 }
//now shortCities has just 'Ashland", cities is unchanged

Passing a function into a function!

pass the transformation into the collection

Great type inference for Function types

there’s “=>” again


Functional Scala

val shortCities = cities.filter { city => city.length < 8 }

more type inference -
scala already knows the
type of “city”, since cities
is a List[String]


Functional Scala

val shortCities = cities.filter { _.length < 8 }

scala uses “_” as the all of these examples
placeholder variable, for compile down to the
when naming the same code
variable doesnt help
its a matter of style
like “it” in groovy when considering how
much inference to use


Functional Scala
Functional style

Imperative style
List<String> shortCities = getShortCityNames(cities);

public List<String> getShortCityNames(List<String cities) {
List<String> shortCityNames = new ArrayList<String>();
for (String city : cities) {
if (city.length() < 8) {
shortCityNames.add(city);
}
}
return shortCityNames;
} This is where bugs live

Functional Scala
Functional style

This is where bugs live

Put the fun back in
functions

val (minors, adults) = people partition (_.age < 18)

make t wo lists out of 1,
based on a predicate

multiple assignment


Put the fun back in
functions

//total salary for all employees in accounting
val totalSalaryForAccountingDept = employees
.filter(_.dept == "Accounting")
.map(_.salary)
.sum

Here, “map” takes a list of
important to note these
are creating new
collections, not

Employee objects, and modifying “employees”

creates a List of salaries


Put the fun back in
functions

var clicks = 0
reactions += {
case ButtonClicked(b) =>
clicks += 1
label.text = "Number of button clicks: "+ clicks
}


Custom control
structures
public void transfer(Account fromAccount,
Account toAccount,
MonetaryAmount amount) { where’s the bugs?
Transaction t = null;
try {
t = acquireTransaction();
fromAccount.debit(amount);
toAccount.credit(amount);
t.commit();
}
catch (Exception e) {
throw new RuntimeException("Failed to transfer ", e);
}
finally {
t.close();
}
}


Custom control
structures
public void transfer(Account fromAccount,
Account toAccount,
MonetaryAmount amount) {
Transaction t = null;
try {
t = acquireTransaction();
fromAccount.debit(amount);
toAccount.credit(amount);
t.commit();
}
catch (Exception e) {
t.rollback();
throw new RuntimeException("Failed to transfer ", e);
}
finally {
if (t != null) {
t.close();
}
}
}


19 lines - where’d my
code go?
def transfer(fromAccount:Account, toAccount:Account, amount:MonetaryAmount) {

transaction {
fromAccount.debit(amount)
toAccount.credit(amount)
}

}


Custom control
structures unit is scala-speak for
void

this is a function that
takes nothing and
def transaction(transactionCode: => Unit) { returns nothing

val transaction = null
try {
transaction = acquireTransaction()
transactionCode
transaction.commit Just pass in the code
} catch {
case e: Exception => that should be in TX
transaction.rollback
throw new RuntimeException("Tx failed: rolled back", e)
}
finally {
if (transaction != null) {
transaction.close
}
}
}


Why ﬁrst-class
functions
• Move boilerplate to library, increase
readability
• Less “book-keeping” and mutable variables
means less bugs
• All the advantages of libraries over
boilerplate: better tuning, quality, better
imagine these operations
abstractions being executed in parallel
on a multicore


I resisted list comprehensions when I first
learned Python, and I resisted filter and map
even longer. I insisted on making my life more
difficult, sticking to the familiar way of for loops
and if statements and step-by-step code-centric
programming. And my Python programs looked a
lot like Visual Basic programs, detailing every
step of every operation in every function. And
they had all the same types of little problems and
obscure bugs. And it was all pointless.

— Mark Pilgrim, Dive into Python, §16.5


Scala is functional

• Immutability/No Side Effects

• Type inference


What’s wrong with side
effects and mutability?
• What do you hate about bad code?
• No seperation of concerns
• Variables changing from many different
paths
• Hard to test or reason about
• Control ﬂow is all over the place

Mutability is surprising
Map<Widget, String> statusForWidget = new HashMap<Widget, String>();
Widget w = new Widget("Whiz", 12);
statusForWidget.put(w, "operational");
w.setName("Whizzer");
System.out.println(statusForWidget.get(w)); //Null!

why doesnt this work?


Mutability has unintended
consequences
private List<Widget> widgets = loadWidgets();
//......
public List<Widget> getAllWidgets() {
return widgets;
}

thing.getAllWidgets().remove(0);

“trust boundaries”

think about someone
downstream calling
“retainAll” on your
widgets field


Mutability is hard to
maintain
public class Account {
private int amount;

public Account() {
}

public String getSummary() {
//can i safely use amount here?
return "Account currently has " + amount + " dollars";
}

public void setAmount(int amount) {
this.amount = amount;
}
}


Side effects can blow
your mind
public class DateParseServlet extends HttpServlet {
private DateFormat dateParser = new SimpleDateFormat("yyyyMMdd");

@Override
protected void doGet(HttpServletRequest req,
HttpServletResponse resp) {
resp.getWriter().print(
dateParser.format(req.getParameter("dateString")));
}
}
Where’s the bug? item #13 in “effective
java”


Why immutable and
side-effect free?

• Easier to reason about
• No need to worry about what callers do
with data they get from you
• We live in a multi-threaded world


Scala and Immutability

• val vs. var
• Collections library
• Actors


val vs. var
class Person(val first:String, val last:String)

• This class is immutable - ﬁrst and last are
read-only
• val does NOT means “immutable”: it means
“non-reassignable” or “ﬁnal”. Your class is
only immutable if it’s “vals all the way
down”


Collections library

• Scala provides two collection libraries:
immutable and mutable
• immutable is “default”


Actors

• Scala preferred concurrent programming
model
• No Runnable, synchronized, no locks, no
shared mutable state - a “developer-
centric” approach
• Modeled from Erlang


Actors
object Accumulator extends Actor { Accumulator.start
def act = { for(i <- (1 to 100)) {
var sum = 0 async {
loop {
react {
Shared state Accumulator ! Accumulate(i)
}
case Accumulate(n) => sum += n }
case Reset => sum = 0
case Total => reply(sum); exit Accumulator !? Total; match {
} case result: Int => println(result)
} }
}
}

Lots of threads hitting the actor
actors are “shared
nothing” -

messages are immutable

mutable state is private

Food for thought

• What effects does immutability have on
performance?
• What effects does immutability have on
correctness?


Scala is functional

• Immutability/Side Effects
• Type inference


Pattern Matching
• Pattern matching is a switch statement
• Matching on properties of object graphs
(not just ints)
• Matching on object type
• Extracting values
• Compiler warning/RuntimeException on
non-match


Pattern Matching
def parseBoolean(a: Any):Boolean = {
val stringTruthies = Set("true", "1", "yes")
a match {
case b:Boolean => b
case i:Int => i == 1
case s:String => stringTruthies.contains(s)
}
}

instanceof/cast
extraction of data
runtime exception if no match

remember: last expression is the return statement


Pattern Matching:
Case Classes
case class Town(city:String, state:String)
case class Person(first:String, last:String, hometown:Town)

def getTownsThatHaveSmithsInVirginia(people:List[Person]) = {
people.collect { person =>
person match {
case Person(_, "Smith", Town(town, "Virginia")) => town
case _ =>
}
}
}

note the _ to say “
any first name”

note the embedded Town object

object graph matching/ type matching/data extraction


Case classes are great
for representing your
“Data objects”:
• Automatically define an “extractor” for
pattern matching
• Automatically define equals and hashCode
based on fields
• Automatically define toString

case classes are great:
copy
val youngy = new Person("Youngy", "McBaby", 2, today)
val youngyTwin = youngy.copy(first = "Youngina")

did you notice named
params?


Express powerful
concepts concisely

case class Town(city:String, state:String)
case class Person(first:String, last:String, hometown:Town)

Converted to Java...


class Person {
public class Town {
public static Person Person(String first, String last, Town homeTown) {
public static Town Town(String city, String state) {
return new Person(first, last, homeTown);
return new Town(city, state);
}
}

private final String city;
private final String first;
private final String state;
private final String last;
private final Town homeTown;
public Town(String city, String state) {
this.city = city;
public Person(String first, String last, Town homeTown) {
this.state = state;
this.first = first;
}
this.last = last;
this.homeTown = homeTown;
public String getCity() {
}
return city;
}
public String getFirst() {
return first;
public String getState() {
}
return state;
}
public String getLast() {
return last;
@Override
}
public int hashCode() {
final int prime = 31;
public Town getHomeTown() {
int result = 1;
return homeTown;
result = prime * result + ((city == null) ? 0 : city.hashCode());
}
result = prime * result + ((state == null) ? 0 : state.hashCode());
return result;
@Override
}
@Override
int result = 1;
public boolean equals(Object obj) {
result = prime * result + ((first == null) ? 0 : first.hashCode());
if (this == obj)
result = prime * result + ((homeTown == null) ? 0 : homeTown.hashCode());
return true;
result = prime * result + ((last == null) ? 0 : last.hashCode());
if (obj == null)
return result;
return false;
}
if (getClass() != obj.getClass())
return false;
@Override
Town other = (Town) obj;
if (city == null) {
if (this == obj)
if (other.city != null)
return true;
return false;
if (obj == null)
}
return false;
else if (!city.equals(other.city))
return false;
return false;
if (state == null) {
Person other = (Person) obj;
if (other.state != null)
if (first == null) {
return false;
if (other.first != null)
}
return false;
else if (!state.equals(other.state))
}
return false;
else if (!first.equals(other.first))
return true;
return false;
}
if (homeTown == null) {
if (other.homeTown != null)
@Override
return false;
public String toString() {
}
return "Town [city=" + city + ", state=" + state + "]";
else if (!homeTown.equals(other.homeTown))
}
return false;
}
if (last == null) {
if (other.last != null)
return false;
}
else if (!last.equals(other.last))
return false;
return true;
}

@Override
return "Person [first=" + first + ", homeTown=" + homeTown + ", last=" + last + "]";
}
}


public class Town {
public static Town Town(String city, String state) {
return new Town(city, state);
}
private final String city;
private final String state;

public Town(String city, String state) {
this.city = city;
this.state = state;
}

public String getCity() {
return city;
}

public String getState() {
return state;
}

@Override
int result = 1;
result = prime * result + ((city == null) ? 0 : city.hashCode());
result = prime * result + ((state == null) ? 0 : state.hashCode());
return result;
}

@Override
if (this == obj)
return true;
if (obj == null)
return false;
return false;
Town other = (Town) obj;
if (city == null) {
if (other.city != null)
return false;
}
else if (!city.equals(other.city))
return false;
if (state == null) {
if (other.state != null)
return false;
}
else if (!state.equals(other.state))
return false;
return true;
}

@Override
return "Town [city=" + city + ", state=" + state + "]";
}
}


Type inference

• Local variables
• Return type
• Type parameters


More noise reduction

• Semicolons
• Dots
• Braces
• Parenthesis
• return


• How does type inference affect readability?
reducing boilerplate is obvious win
but what about inherited/old code?
scala style guide recommends never inferring return type


Extra innings

• Option
• Implicit conversions
• Named parameters/default values


Where’s the bug?

String name = getPerson("person1234").getName();


Where’s the bug?

String name = getPerson("person1234").getName();

Null!


How do you know
when a value is null?
• Great guesser • Use a language that
supports nullable tyypes
• Clairvoyance
• Documentation
• Naming conventions
• Sentinel values
• Just pretend nothing will
be null • Great tester

• Just pretend everything
might be null

• Annotations


How do you know
when a value is null?


Scala’s approach:
Option[T]
def getPerson(id:String):Option[Person] = {
if (id == "person123") {
Some(new Person("Bob"))
}
else {
None
}
}


Scala’s approach:
Option[T]
• Option is extended by Some and None
• Encodes optional values into your API with
zero magic - can be used in Java
• Scala’s type inference keeps the noise to a
minimum
• May seem like an extra burden, but the
burden is already there once you have
optional values

Scala’s approach:
Option[T]
//previous approach: doesnt compile!
val name = getPerson("person123").name

//.get will throw a RuntimeException if it is null here
val name2 = getPerson("person123").get.name

//get the value, or use a default
val name3 = getPerson("person123").getOrElse("Name unknown")

//in some cases (not this one), pattern matching is nice
val name4 = getPerson("person123") match {
case Some(name) => name
case None => "Name Unknown"
}


Implicit conversions
• Automatically convert types
• Commonly used to add behavior to
existing closed source classes
• Programmer controls the scope of the
conversions - not global
• Goes beyond just adding methods


Implicit Conversions
A Rich Wrapper
class RichInt(i:Int) {
def times(f: => Unit) = {
for (x <- 0 to i) f
}
}

new RichInt(5).times { println("hey!") }

yuck


Implicit Conversions
class RichInt(i:Int) {
def times(f: => Unit) = {
for (x <- 0 to i) f
}
}

implicit def intToRichInt(i:Int) = new RichInt(i)

5 times println("hey!")


Named and default
parameters
def sendEmail(
to : List[String],
from : String,
subject : String = "No subject",
cc : List[String] = Nil,
bcc : List[String] = Nil,
attachments : List[File] = Nil
)


Named and default
parameters
sendEmail(List("a@a.com"), "b@b.com")

sendEmail(to = List("a@a.com"),
from = "b@b.com",
subject ="Status Check!")

sendEmail(to = List("a@a.com"),
from = "b@b.com",
subject ="Status Check!",
attachments = List(new File("coolpic.jpg")))


Out of time
• DSLs • Currying

• Sane generics (no
wildcards!)

• Non-alpha method
names

• Multi-line strings

• Structural typing

• XML Literals


My top10 Aspects of
Scala
• Java Interop • rich collections library

• Concise/Type inference • default/named params

• Functions-as-objects • case classes

• Some/None/Option • REPL

• implicit conversions

• traits


These concepts are in
other statically typed
JVM languages too!
• Fantom (from Richmond!)
• Gosu
• Java 8 (currently “late 2012”)


Go try it!

• http://scala-lang.org
• http://www.simplyscala.com/ - try simple
samples without installing
• I will be posting slides + a pre-built Eclipse
with this talks code on my blog.


Thanks!
adamrabung@gmail.com
http://squirrelsewer.blogspot.com
twitter:squirrelsewer


Introduction to Scala for Java Programmers

More Related Content

What's hot

Viewers also liked

Similar to Introduction to Scala for Java Programmers

Recently uploaded

Introduction to Scala for Java Programmers