• Stars
    star
    106
  • Rank 325,871 (Top 7 %)
  • Language
    Scala
  • Created over 3 years ago
  • Updated 11 months ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Monadic Reflection

This project provides support for monadic reflection (Filinski 1994, 1999) to integrate monadic code with direct style code.

Tired of writing code using for-comprehensions?

The monadic-reflection library provides a convenient alternative!

Before

for {
  _ <- monadicActionA()
  r <- monadicActionB()
  result <- if (predicate(r)) {
    monadicActionC()
  } else {
    monadicActionD()
  }
} yield result

After

effectfulActionA()
if (predicate(effectfulActionB())) {
  effectfulActionC()
} else {
  effectfulActionD()
}

Looks familiar? Yes, it is just the direct-style code you would write in an imperative programming language.

Concepts

The underlying idea is very simple: Instead of using your monadic type constructor M[A] everywhere, your effectful programs now have the type CanReflect[M] ?=> A where CanReflect is a type defined by the monadic-reflection library.

As you can see from the type, given the capability CanReflect[M] you immediately get a value of type A that you can just use in direct-style. No need for flatMap and friends.

The best thing is, that you can go back and forth between the two representations:

trait Monadic[M[_]] {
  // embed a monadic value into direct style
  def reflect[R](mr: M[R])(using r: CanReflect[M]): R = r.reflect(mr)

  // reveal the monadic structure of a direct-style program
  def reify[R](prog: CanReflect[M] ?=> R): M[R]
}

How can I use this with my monad?

All you need to do is implement the Monadic trait which has two abstract methods:

def pure[A](a: A): M[A]
def sequence[X, R](init: M[X])(f: X => Either[M[X], M[R]]): M[R]

The first should look very familiar to you -- and if you already have a monad is very easy to implement. The second is just a slight variation of flatMap. In order to be stack safe you need to make sure to either implement sequence as a tail recursive function, or perform trampolining on your own.

Well, there is a fineprint: You also need to run your programs using a special JDK fork called "Project Loom". See below for more details.

Example Integrations

We provide a few case studies showing how to program with established monadic libraries in direct style:

Dependencies

To implement monadic reflection we require some implementation of (delimited) continuations. At the moment, our library only runs on JDK >= 21

Run Sbt

Finally, since we are accessing jvm internal types (Continuation and ContinuationScope), we need to allow our program to access them. This is done by forking the process in the sbt configuration.

If this does not work for you (for whatever reason), you can try to run sbt with:

sbt -J--add-exports=java.base/jdk.internal.vm=ALL-UNNAMED

Some experimental performance optimizations of project loom can be enabled by

-XX:-DetectLocksInCompiledFrames -XX:+UnlockDiagnosticVMOptions -XX:+UnlockExperimentalVMOptions -XX:+UseNewCode

More Repositories