Practical effect composition library based on abstract wrapping type and the free monad
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
dsl/src/main/scala/freedsl/dsl
errorhandler/src/main/scala/freedsl/errorhandler
example/src/main/scala/freedsl/example
filesystem/src/main/scala/freedsl/filesystem
io/src/main/scala/freedsl/io
log/src/main/scala/freedsl/log
project
random/src
system/src/main/scala/freedsl/system
tool/src
.gitignore
README.md
build.sbt
version.sbt

README.md

FreeDSL

Overview

FreeDSL is a library for pure composition of side effects, weaving typeclasses on an abstract wrapping type and the Free Monad.

Examples

Creating a new DSL

import freedsl.dsl._

object Random {

  def interpreter(random: util.Random) = new Interpreter {
    def nextDouble(implicit context: Context) = random.nextDouble
    def nextInt(n: Int)(implicit context: Context) = random.nextInt(n)
    def shuffle[A](s: Seq[A])(implicit context: Context) = result(random.shuffle(s))
    def fakeError(implicit context: Context) = failure(FakeError("for some reason there was an error"))
  }

  def interpreter(seed: Long): Interpreter = interpreter(new util.Random(seed))
  
  case class FakeError(cause: String) extends Error

}

@dsl trait Random[M[_]] {
  def nextDouble: M[Double]
  def nextInt(n: Int): M[Int]
  def shuffle[A](s: Seq[A]): M[Seq[A]]
  def fakeError: M[Unit]
}

Using your new DSL

Merging interpreters

import cats._
import cats.implicits._
import freedsl._
import freedsl.random._
import freedsl.util._
import freedsl.log._

import squants.time.TimeConversions._

// Pure functions that describe computations depending on side effects
def randomData[M[_]](implicit randomM: Random[M]): M[Seq[Int]] =
  randomM.shuffle(Seq(1, 2, 2, 3, 3, 3))

def randomSleep[M[_]: Monad](implicit randomM: Random[M], utilM: Util[M], logM: Log[M]): M[Unit] = for {
  s <- randomM.nextInt(10)
  _ <- logM.print(s"Sleeping for $s seconds")
  _ <- utilM.sleep(s seconds)
} yield ()


// Construct the interpreter and a type M along with implicit instances of Random[M], Util[M] and Log[M]
// they are build using the free monad and the freek library

val interpreter = merge(Util.interpreter, Random.interpreter(42), Log.interpreter)
import interpreter.implicits._

def prg =
  for {
    b  randomData[interpreter.M]
    _  randomSleep[interpreter.M]
  } yield b
  

// All the side effects take place here in the interpreter
interpreter.run(prg) match {
  case Right(v) => println(s"This is a success: $v")
  case Left(e) => println(s"OhOh, error: $e")
}

Merging DSLs

Types can be merged independently from intepreters. For instance:

val merged = merge(Util, Random, Log)
import merged.implicits._

def prg =
  for {
    b  randomData[merged.M]
    _  randomSleep[merged.M]
  } yield b

  
val interpreter = merge(Util.interpreter, Random.interpreter(42), Log.interpreter)

// All the side effects take place here in the interpreter
interpreter.run(prg) match {
  case Right(v) => println(s"This is a success: $v")
  case Left(e) => println(s"OhOh, error: $e")
}

Mutli-level merging

DSLs and interpreters can be merged at multiple scala. For instance:

val partialMerge = merge(Util, Random)
val merged = merge(partialMerge, Log)
import merged.implicits._

def prg =
  for {
    b  randomData[merged.M]
    _  randomSleep[merged.M]
  } yield b

  
val interpreter = merge(Util.interpreter, Random.interpreter(42), Log.interpreter)

// All the side effects take place here in the interpreter
interpreter.run(prg) match {
  case Right(v) => println(s"This is a success: $v")
  case Left(e) => println(s"OhOh, error: $e")
}

And for interpreters:

// Construct the interpreter and a type M along with implicit instances of Random[M], Util[M] and Log[M]
// they are build using the free monad and the freek library

val partialMerge = merge(Util.interpreter, Random.interpreter(42))

// In case several interpreters are provided for a given DSL, the first one in the list is retained
val interpreter = merge(partialMerge, Log.interpreter)
import interpreter.implicits._

def prg =
  for {
    b  randomData[interpreter.M]
    _  randomSleep[interpreter.M]
  } yield b
  

// All the side effects take place here in the interpreter
interpreter.run(prg) match {
  case Right(v) => println(s"This is a success: $v")
  case Left(e) => println(s"OhOh, error: $e")
}

Wrapping errors

To return meaningful errors, it is sometimes necessary to wrap errors produced by side effects:

  object DSLTest1M {
    def interpreter = new Interpreter {
      def get(implicit context: Context) = failure(DSL1Error("Boooh"))
    }

    case class DSL1Error(s: String) extends Error
  }

  @dsl trait DSLTest1M[M[_]] {
    def get: M[String]
  }

  case class WrapError(e: Error) extends Error

  def prg[M[_]](implicit dSLTest1M: DSLTest1M[M], errorWrapping: ErrorWrapping[M]) = errorWrapping.wrap(e => WrapError(e)) {
    dSLTest1M.get
  }

Getting FreeDSL

FreeDSL is published to sonatype.

FreeDSL supports Scala 2.11 and 2.12. If you use SBT, you can include FreeDSL via the following build.sbt snippets:

// Repository for Freek
resolvers += Resolver.bintrayRepo("projectseptemberinc", "maven")

// For scala 2.12, for 2.11 use Miles Sabin's plugin for type unification.
scalacOptions := Seq("-Ypartial-unification")

def freedslVersion = "0.9-SNAPSHOT"

// pick a particular subproject
libraryDependencies += "fr.iscpif.freedsl" %% "util" % freedslVersion,
libraryDependencies += "fr.iscpif.freedsl" %% "random" % freedslVersion % "test"

Detailed Information

TODO

Known Issues

DSLs don't suport default parameter value. It raises an error at runtime, such as:

java.lang.AbstractMethodError: freedsl.dsl.DSLTest$InterpretationContext$1$MergedDSL$1$$anon$12$$anon$10.defaultValue$default$2()Ljava/lang/String;

Future Work

TODO

Copyright and License

All code is available to you under the LGPL license, available at https://www.gnu.org/licenses/lgpl.html and also in the COPYING file.

Copyright Romain Reuillon, 2016

No Warranty

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.