Smoothing Your Java with DSLs

Smoothing Your Java with DSLs Dmit ry Jemerov Development Lead JetBrains, Inc. ht tp://www.jetbrains.com/

Learn about the multiple ways you can use domain-specific languages in your Java project to make your code more readable and maintainable

Agenda DSL Overview DSLs in Java DSLs in Scripting Languages JetBrains MPS

What is a DSL? Computer programming language Limited expressiveness Domain focus

DSL Classification External Regular Expressions SQL CSS

DSL Classification External XML-based Ant XSLT

DSL Classification External XML-based Internal Query integration: Squill Testing: JMock, EasyMock

DSL Classification External XML-based Internal Scripting Build: Gradle, Buildr Testing: easyb UI: Glimmer, Monkeybars

DSL Classification External XML-based Internal Scripting Language Workbench MPS “Oslo”

Why Use DSLs? Raising level of abstraction Separating the “what” from the “how” Separating stable and frequently changing parts of the system Enabling domain expert contributions

DSLs: Code or Data? Depends on where you look from TrueType fonts: actually code for a rather powerful VM

Integrating DSLs Interpretation Model is built by executing internal DSL or parsing external DSL Interpreter executes operations described by DSL Code Generation DSL drives generation of host language code DSL-generated code compiled with rest of app

Squill squill .from(pets) .where(eq(pets.species, “Cat”)) .orderBy(asc(pets.name)) .selectList(pets.name, pets.birthDate)); https://squill.dev.java.net/

EasyMock expect(mock.voteForRemoval(“A")) .andReturn((byte) 42) .atLeastOnce(); http://www.easymock.org/

Regular vs Fluent API Regular: a method call executes an operation object.verbNoun(param1, …, paramN) Fluent: a method call tells one thing about an operation that will be executed object.noun1(param1).noun2(param2)…

Fluent API Patterns Method chaining Builder objects Static factory methods Static imports Initializer blocks in anonymous classes

Method Chaining class C { C value1(Object v1) { this.value1 = v1; return this; } C value2(Object v2) { … } } c.value1(o1).value2(o2)

Builder Objects class CBuilder { CBuilder value1(Object v1) { … } CBuilder value2(Object v2) { … } C build() { return new C(value1, value2); } }

Factory Methods, Static Imports class CFactory { static <T> C<T> newC() { return new C<T>(); } } import static CFactory.*; C<String> c = newC();

Initializer Blocks (JMock) context.checking(new Expectations() {{ oneOf (subscriber).receive(message); }});

Java Internal DSL Advantages Low entry barrier, no external dependencies Full IDE support (code completion etc.) Strong typing and compile-time error checking

Java Internal DSL Drawbacks Readability is not as good as with other solutions

easyb given "a stack with one pushed value", { stack = new Stack() pushVal = "foo" stack.push(pushVal) } when "pop is called", { popVal = stack.pop() } then "that object should be returned", { popVal.shouldBe pushVal } http://easyb.org/

Glimmer shell { text "Contact Manager“ composite { layout GridLayout.new(2, false) label { text "First &Name: "} text { text bind(presenter, :first_name) } } } http://rubyforge.org/projects/glimmer/

RubyMine Parameter Definitions define_params 'validates_presence_of' , migration_ref , [migration_ref , { :message => nil , :on => one_of(:save , :create , :update) , :if => method_ref( 'ActiveRecord::Base' ) }]

Scripting Language Advantages Closures Properties List/map literals Operator overloading Handling of missing methods

Integrating DSL Scripts Joint compilation Script statically compiled to bytecode Allows direct calls of script code from Java Dynamic invocation Script invoked via javax.scripting or language-specific embedding API Allows dynamic discovery of scripts Allows modifying scripts without app restart or recompilation

MPS History Started as a research project back in 2004 JetBrains issue tracker project being developed entirely in MPS Web framework Data access framework End-user functionality Released as open-source public beta in December 2008

Key Concepts of MPS Language defines a set of concepts and relationships between them Models are instances of a particular concept Editors define how models of a particular concept are represented in the MPS IDE Generators define how models are translated into a different language

What MPS doesn’t have No text representation for code in a particular language No grammar for languages No way to import code in existing text-based language No dependency on specific output language generated Could be Java, JavaScript, CSS, XML…

Advanced MPS Reusing languages Business rules need formulas Can reuse MPS expression language: editor, generation etc. Extending languages Closures, ‘using’ blocks, complex numbers…

Summary You can use the fluent programming style in Java to make your code easier to read and less verbose DSLs built with scripting languages like Groovy and JRuby bring even further readability benefits JetBrains MPS is a whole new approach to DSLs

For More Information Martin Fowler’s DSL book: http://martinfowler.com/dslwip/ JetBrains MPS: http://www.jetbrains.com/mps/

Q&A Dmitry Jemerov [email_address]

Smoothing Your Java with DSLs

More Related Content

What's hot (20)

Viewers also liked (10)

Similar to Smoothing Your Java with DSLs (20)

More from intelliyole (7)

Recently uploaded (20)

Smoothing Your Java with DSLs