Search code examples
groovydesign-patternsabstract-syntax-treedelegationmixins

Difference between @Delegate and @Mixin AST transformations in Groovy


What's the difference between @Delegate and @Mixin AST transformations in Groovy.

Maybe my question has to do with OO and when apply different patterns, but I use both and I can achieve the same behavior.

class Person {
    String name = "Clark"
    def walk() { "Walk" }
}

@Mixin(Person)
class Superhero {
    def fly() { "Fly" }
}

def superman = new Superhero()
assert superman.name == "Clark"
assert superman.walk() == "Walk"
assert superman.fly() == "Fly"

class Person {
    String name = "Clark"
    def walk() { "Walk" }
}

class Superhero {
    @Delegate Person person
    def fly() { "Fly" }
}

def superman = new Superhero(person: new Person())
assert superman.name == "Clark"
assert superman.walk() == "Walk"
assert superman.fly() == "Fly"

Solution

  • The behavior is similar, but @Delegate and @Mixin are implemented completely differently.

    @Delegate generates accessor methods at compile time. Superhero will have a method called walk() that simply calls person.walk(). The generated methods can be seen by dumping the Superhero class file with javap.

    @Mixin, on the other hand just creates a small stub that mixs in the Person methods at runtime. It uses groovy's meta-object protocol to allow Superhero to respond to Person's methods. In this case, you won't see any Person methods in Superhero.class.

    @Delegate has the advantage that the methods are callable from Java and it avoids doing a dynamic invocation. In addition, @Mixin can't augment the class with properties.