Scala Talk at FOSDEM 2009

A Scalable Language Martin Odersky FOSDEM 2009

The software landscape today … … resembles a tower of Babel with many little (or not so little) languages playing together. E.g. JavaScript on the client Perl/Python/Ruby/Groovy for server side scripting JavaFX for the UI Java for the business logic SQL for database access all cobbled together with a generous helping of XML.

This is both good and bad Good : Every language can concentrate on what it’s best at. Bad : Cross language communication: complicated, fragile, source of misunderstandings. Problematic : Cross language communication is controlled by a common type system (neither static nor dynamic). It's based on low-level representations such as XML trees or (worse) strings (as in JDBC database queries).

Alternative: Scalable languages A language is scalable if it is suitable for very small as well as very large programs. A single language for extension scripts and the heavy lifting. Application-specific needs are handled through libraries and embedded DSL's instead of external languages. Scala shows that this is possible.

Scala is a scripting language It has an interactive read-eval-print-loop (REPL). Types can be inferred. Boilerplate is scrapped. scala> var capital = Map("US"  "Washington", "France"  "Paris") capital: Map[String, String] = Map(US  Washington, France  Paris) scala> capital += ("Japan"  "Tokio") scala> capital("France") res7: String = Paris

Scala is the Java of the future It has basically everything Java has now. (sometimes in different form) It has closures. (proposed for Java 7, but rejected) It has traits and pattern matching. (I would not be surprised to see them in Java 8, 9 or 10) It compiles to .class files, is completely interoperable and runs about as fast as Java object App { def main ( args : Array [ String ]) { if ( args exists (_. toLowerCase == "-help" )) printUsage () else process ( args ) } }

Interoperability Scala fits seamlessly into a Java environment Can call Java methods, select Java fields, inherit Java classes, implement Java interfaces, etc. None of this requires glue code or interface descriptions Java code can also easily call into Scala code Scala code resembling Java is translated into virtually the same bytecodes.  Performance is usually on a par with Java

Scala is a composition language New approach to module systems: component = class or trait composition via mixins Abstraction through parameters, abstract members (both types and values), self types gives dependency injection for free trait Analyzer { this : Backend => … } trait Backend extends Analyzer with Optimization with Generation { val global : Main import global ._ type OutputMedium <: Writable }

Is Scala a “kitchen-sink language”? Not at all. In terms of feature count, Scala is roughly comparable to today’s Java and smaller than C# or C++. But Scala is deep , where other languages are broad . Two principles: 1. Focus on abstraction and composition, so that users can implement their own specialized features as needed. 2. Have the same sort of constructs work for very small as well as very large programs.

Scala compared to Java Modeled in libraries: assert, enums, properties, events, actors, using, queries, … raw types + abstract types Scala removes Scala adds - wildcards + existential types enums + pattern matching - special treatment of interfaces + mixin composition with traits - break, continue + closures - primitive types + operator overloading - static members + a pure object system

Scala cheat sheet (1): Definitions Scala method definitions: def fun(x: Int): Int = { result } def fun = result Scala variable definitions: var x: Int = expression val x: String = expression Java method definition: int fun( int x) { return result } (no parameterless methods) Java variable definitions: int x = expression final String x = expression

Scala cheat sheet (2): Expressions Scala method calls: obj.meth(arg) obj meth arg Scala choice expressions: if (cond) expr1 else expr2 expr match { case pat 1 => expr 1 .... case pat n => expr n } Java method call: obj.meth(arg) (no operator overloading) Java choice expressions, stmts: cond ? expr1 : expr2 if (cond) return expr1; else return expr2; switch (expr) { case pat 1 : return expr 1 ; ... case pat n : return expr n ; } // statement only

Scala cheat sheet (3): Objects and Classes Scala Class and Object class Sample(x: Int, val p: Int) { def instMeth(y: Int) = x + y } object Sample { def staticMeth(x: Int, y: Int) = x * y } Java Class with statics class Sample { private final int x; public final int p; Sample( int x, int p) { this.x = x; this.p = p; } int instMeth( int y) { return x + y; } static int staticMeth( int x, int y) { return x * y; } }

Scala cheat sheet (4): Traits Scala Trait trait T { def abstractMth(x: String): Int def concreteMth(x: String) = x + field var field = “!” } Scala mixin composition: class C extends Super with T Java Interface interface T { int abstractMth(String x) } (no concrete methods) (no fields) Java extension + implementation: class C extends Super implements T

Spring Cleaning Scala’s syntax is lightweight and concise. Due to: semicolon inference, type inference, lightweight classes, extensible API’s, closures as control abstractions. Average reduction in LOC: ≥ 2 due to concise syntax and better abstraction capabilities  Scala feels like a cleaned up Java … var capital = Map ( "US" -> "Washington" , "Canada" -> "ottawa" ) capital += ( "Japan" -> "Tokyo" ) for (c <- capital.keys) capital(c) = capital(c).capitalize assert(capital( "Canada" ) == "Ottawa" )

… with one major difference It's x: Int instead of int x Why the change? Works better with type inference: var x = 0 instead of x = 0 // that's not a definition! Works better for large type expressions: val x: HashMap[String, (String, List[Char])] = … instead of public final HashMap<String, Pair<String, List<Char>>> x = …

Scalability demands extensibility Take numeric data types Today's languages support int, long, float, double . Should they also support BigInt, BigDecimal, Complex, Rational, Interval, Polynomial ? There are good reasons for each of these types But a language combining them all would be too complex. Better alternative: Let users grow their language according to their needs.

Adding new datatypes - seamlessly For instance type BigInt : def factorial ( x : BigInt ): BigInt = if ( x == 0) 1 else x * factorial ( x - 1) Compare with using Java's class: import java.math.BigInteger def factorial ( x : BigInteger ): BigInteger = if ( x == BigInteger.ZERO) BigInteger . ONE else x . multiply (factorial(x.subtract(BigInteger.ONE))) }

Implementing new datatypes - seamlessly Here's how BigInt is implemented import java.math.BigInteger class BigInt ( val bigInteger : BigInteger ) extends java . lang . Number { def + ( that : BigInt ) = new BigInt ( this . bigInteger add that . bigInteger ) def - ( that : BigInt ) = new BigInt ( this . bigInteger subtract that . bigInteger ) … // other methods implemented analogously } + is an identifier; can be used as a method name Infix operations are method calls: a + b is the same as a.+(b) a add b is the same as a.add(b)

Adding new control structures For instance using for resource control (proposed for Java 7) Instead of: using ( new BufferedReader ( new FileReader ( path ))) { f => println ( f . readLine ()) } val f = new BufferedReader ( new FileReader ( path )) try { println ( f . readLine ()) } finally { if ( f != null ) f . close () }

Implementing new control structures: Here's how one would go about implementing using : def using [ T <: { def close () }] ( resource : T ) ( block : T => Unit ) { try { block ( resource ) } finally { if ( resource != null ) resource . close () } } T is a type parameter... … supporting a close method A closure that takes a T parameter

Break and continue Scala does not have them. Why? They are a bit imperative; better use many smaller functions. Issues how to interact with closures. They are not needed! We can support them purely in the libraries. import scala.util.control.Breaks._ breakable { for (x <- elems) { println(x * 2) if (x > 0) break } }

Getting back break and continue

What makes Scala scalable? Many factors: strong typing, inference, little boilerplate,… But mainly, its tight integration of functional and object-oriented programming Functional programming: Makes it easy to build interesting things from simple parts, using higher-order functions, algebraic types and pattern matching, parametric polymorphism. Object-oriented programming: Makes it easy to adapt and extend complex systems, using subtyping and inheritance, dynamic configurations, classes as partial abstractions.

Scala is object-oriented Every value is an object Every operation is a method call Exceptions to these rules in Java (such as primitive types, statics) are eliminated. scala> (1).hashCode res8: Int = 1 scala> (1).+(2) res10: Int = 3

Scala is functional Scala is a functional language, in the sense that every function is a value. Functions can be anonymous, curried, nested. Many useful higher-order functions are implemented as methods of Scala classes. E.g: scala> val matrix = Array(Array(1, 0, 0), | Array(0, 1, 0), | Array(0, 0, 1)) matrix: Array[Array[Int]] = Array([I@164da25,… scala> matrix.exists(row => row.forall(0 ==)) res13: Boolean = false

Functions are objects If functions are values, and values are objects, it follows that functions themselves are objects. The function type S => T is equivalent to scala.Function1[S, T], where Function1 is defined as follows: So functions are interpreted as objects with apply methods. For example, the anonymous successor function (x: Int ) => x + 1 is expanded to: trait Function1 [ -S , +T ] { def apply ( x : S ): T } new Function1 [ Int , Int ] { def apply ( x : Int ) = x + 1 }

Why should I care? Since ( => ) is a class, it can be subclassed. So one can specialize the concept of a function. An obvious use is for arrays, which are mutable functions over integer ranges. A bit of syntactic sugaring lets one write: a(i) = a(i) + 2 for a.update(i, a.apply(i) + 2) class Array [ T ] ( l : Int ) extends ( Int => T ) { def length : Int = l def apply ( i : Int ): T = … def update ( i : Int , x : T ): Unit def elements : Iterator [ T ] def exists ( p : T => Boolean ) … }

Partial functions Another useful abstraction are partial functions. These are functions that are defined only in some part of their domain. What's more, one can inquire with the isDefinedAt method whether a partial function is defined for a given value. Scala treats blocks of pattern matching cases as instances of partial functions. This lets one write control structures that are not easily expressible otherwise. trait PartialFunction [- A , + B ] extends ( A => B ) { def isDefinedAt ( x : A ): Boolean }

Developing new paradigms Scala's flexibility makes it possible for users to grow the language into completely new paradigms. Case in point: concurrent programming Since Scala is interoperable, Java threads and concurrent libraries are available. But it's also possible to explore completely new paradigms.

Erlang-style actors Two principal constructs (adopted from Erlang): Send ( ! ) is asynchronous; messages are buffered in an actor's mailbox. receive picks the first message in the mailbox which matches any of the patterns msgpat i . If no pattern matches, the actor suspends. // asynchronous message send actor ! message // message receive receive { case msgpat 1 => action 1 … case msgpat n => action n } A pattern matching block of type PartialFunction[ MessageType , ActionType ]

A simple actor case class Data ( bytes : Array [ Byte ]) case class Sum ( receiver : Actor ) val checkSumCalculator = actor { var sum = 0 loop { receive { case Data ( bs ) => sum += hash ( bs ) case Sum ( receiver ) => receiver ! sum } } } } repeatedly receive messages Spawn a new actor

Implementing receive Using partial functions, it is straightforward to implement receive: Here, self designates the currently executing actor, mailBox is its queue of pending messages, and extractFirst extracts first queue element matching given predicate. def receive [ T ] ( f : PartialFunction[Message , T ]): T = { self . mailBox . extractFirst ( f . isDefinedAt ) match { case Some(msg ) => f ( msg ) case None => self . wait ( messageSent ) }}

Other Approaches to Scalability C++ Hard to scale down. Scaling up is possible for expert users. .NET Many languages with common interoperability. Hard to do something that's really different. Java Lingua franca makes it easy to understand other people's code. Not easy to scale down or up  pressure to add new languages.

Where are we now? Scala Easy to scale down and up. Works well with a mix of expert users (for the framework) and non-experts (for the application code). Scala solves the expressiveness challenge for doing this. But does it also solve the safety issues? Problem: How to ensure that domain-specific code stays within its domain-specific library/language? For instance: How to ensure that a query formulated in Scala is non-recursive? Addressed by ongoing project: Pluggable type systems

The Scala community 50000 downloads in 2008 300+ trak contributors 20+ messages/day on the mailing lists Industrial adoption has started, among others at: Twitter, Sony Pictures, Nature.com, Reaktor, Mimesis Republic, EDF Trading, … Scala LiftOff conference, May 2008. Scala talks in many conferences; next two at QCon, London, March 10-12.

Tool support Standalone compiler: scalac Fast background compiler: fsc Interactive interpreter shell and script runner: scala Web framework: lift Testing frameworks: Specs, ScalaCheck, ScalaTest, SUnit, … IDE plugins for: Eclipse (supported by EDF) IntelliJ (supported by JetBrains) Netbeans (supported by Sun)

Who’s using it? Open source projects: lift wicket NetLogo SPDE: Scala branch for Processing Isabelle: GUI and code extractor Companies: Twitter: infrastructure Sony Pictures: middleware Nature.com: infrastructure SAP community: ESME company messaging Reaktor: many different projects Mimesis Republic: multiplayer games EDF: trading, …

Learning Scala To get started: First steps in Scala, by Bill Venners published in Scalazine at www.artima.com Scala for Java Refugees by Daniel Spiewack (great blog series) To continue: Programming in Scala, by Odersky, Spoon, Venners, published by Artima,com Other books are in the pipeline.

Thank You To try it out: scala-lang.org Thanks also to the (past and present) members of the Scala team: Philippe Altherr, Vincent Cremet, Iulian Dragos, Gilles Dubochet, Burak Emir, Sebastian Hack, Philipp Haller, Sean McDirmid, Ingo Meier, Adriaan Moors, Stéphane Micheloud, Nikolay Mihaylov, Anders Nielssen, Tiark Rompf, Lukas Rytz, Michel Schinz, Lex Spoon, Erik Stenman, Geoffrey Alan Washburn, Matthias Zenger .

Relationship between Scala and other languages Main influences on the Scala design: Java, C# for their syntax, basic types, and class libraries, Smalltalk for its uniform object model, Eiffel for its uniform access principle, Beta for systematic nesting, ML, Haskell for many of the functional aspects. OCaml, OHaskel, PLT-Scheme, as other (less tightly integrated) combinations of FP and OOP. Pizza, Multi Java, Nice as other extensions of the Java platform with functional ideas. (Too many influences in details to list them all) Scala also seems to influence other new language designs, see for instance the closures and comprehensions in LINQ/C# 3.0.

Scala Talk at FOSDEM 2009

More Related Content

Scala Talk at FOSDEM 2009