Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Freestyle + Freestyle RPC Introduction

Freestyle + Freestyle RPC Introduction

# Freestyle

This presentation shows the current Features available as of Freestyle 0.4.0 Including Freestyle RPC, a new module to build purely functional microservices.

Freestyle is a library that empowers stack-safe purely functional applications and libraries that support parallel and sequential computations where declaration is decoupled from interpretation.
Freestyle encourages programs built atop Free algebras that are interpreted at the edge of your application ensuring effects are localized and performed in a controlled environment.
Applications built with Freestyle can be interpreted to any runtime semantics supported by the interpreter target type.

http://frees.io
https://github.com/frees-io/freestyle-rpc
https://github.com/47deg/freestyle-intro

Raúl Raja Martínez

November 02, 2017
Tweet

More Decks by Raúl Raja Martínez

Other Decks in Programming

Transcript

  1. Getting into typed FP is hard because No previous CT

    knowledge or math foundations Leaving styles one is used to (ex. OOP) Lack of docs on how to properly use MTL/Tagless and other techniques required for concise FP. Rapid changing ecosystem Scala's encoding Lack of good examples where FP is applied 2
  2. Freestyle's Goals Approachable to newcomers Stack-safe Dead simple integrations with

    Scala's library ecosystem Help you build pure FP apps, libs & micro- services Pragmatism 3
  3. In this talk Freestyle programming style @free, @tagless, @module, @service

    Effects Integrations Optimizations (iota Coproduct, stack-safe @tagless) RPC based Microservices 4
  4. Interface + Impl driven design @free trait Interact { def

    ask(prompt: String): FS[String] def tell(msg: String): FS[Unit] } implicit val handler: Interact.Handler[Future] = new Interact.Handler[Future] { def ask(prompt: String): Future[String] = ??? def tell(msg: String): Future[Unit] = ??? } 5
  5. Boilerplate Reduction > Declaration + @free trait Interact { +

    def ask(prompt: String): FS[String] + def tell(msg: String): FS[Unit] + } - sealed trait Interact[A] - case class Ask(prompt: String) extends Interact[String] - case class Tell(msg: String) extends Interact[Unit] - - class Interacts[F[_]](implicit I: InjectK[Interact, F]) { - def tell(msg: String): Free[F, Unit] = Free.inject[Interact, F](Tell(msg)) - def ask(prompt: String): Free[F, String] = Free.inject[Interact, F](Ask(prompt)) - } - - object Interacts { - implicit def interacts[F[_]](implicit I: InjectK[Interact, F]): Interacts[F] = new Interacts[F] - def apply[F[_]](implicit ev: Interacts[F]): Interacts[F] = ev - } 6
  6. Boilerplate Reduction > Composition + @module trait App { +

    val exerciseOp: ExerciseOp + val userOp: UserOp + val userProgressOp: UserProgressOp + val githubOp: GithubOp + } - type C01[A] = Coproduct[ExerciseOp, UserOp, A] - type C02[A] = Coproduct[UserProgressOp, C01, A] - type ExercisesApp[A] = Coproduct[GithubOp, C02, A] - val exerciseAndUserInterpreter: C01 ~> M = exerciseOpsInterpreter or userOpsInterpreter - val userAndUserProgressInterpreter: C02 ~> M = userProgressOpsInterpreter or exerciseAndUserInterpreter - val allInterpreters: ExercisesApp ~> M = githubOpsInterpreter or userAndUserProgressInterpreter 7
  7. Freestyle's Workflow Declare your algebras Group them into modules Compose

    your programs Provide implicit implementations of each algebra Handler Run your programs at the edge of the world 8
  8. Freestyle's Workflow Declare your algebras import freestyle._ import freestyle.tagless._ object

    algebras { /* Handles user interaction */ @free trait Interact { def ask(prompt: String): FS[String] def tell(msg: String): FS[Unit] } /* Validates user input */ @tagless trait Validation { def minSize(s: String, n: Int): FS[Boolean] def hasNumber(s: String): FS[Boolean] } } 9
  9. Freestyle's Workflow Group them into modules import algebras._ import freestyle.effects.error._

    import freestyle.effects.error.implicits._ import freestyle.effects.state val st = state[List[String]] import st.implicits._ object modules { @module trait App { val validation: Validation.StackSafe val interact: Interact val errorM : ErrorM val persistence: st.StateM } } 10
  10. Freestyle's Workflow Declare and compose programs inside @free, @tagless or

    @module import cats.implicits._ object modules { @module trait App { val validation: Validation.StackSafe val interact: Interact val errorM : ErrorM val persistence: st.StateM def program: FS.Seq[Unit] = for { cat <- interact.ask("What's the kitty's name?") isValid <- (validation.minSize(cat, 5), validation.hasNumber(cat)).mapN(_ && _) //may run ops in parallel _ <- if (isValid) persistence.modify(cat :: _) else errorM.error(new RuntimeException("invalid name!")) cats <- persistence.get _ <- interact.tell(cats.toString) } yield () } } 11
  11. Freestyle's Workflow Declare and compose programs anywhere else def program[F[_]]

    (implicit I: Interact[F], R: st.StateM[F], E: ErrorM[F], V: Validation.StackSafe[F]): FreeS[F, Unit] = { for { cat <- I.ask("What's the kitty's name?") isValid <- (V.minSize(cat, 5), V.hasNumber(cat)).mapN(_ && _) //may run ops in parallel _ <- if (isValid) R.modify(cat :: _) else E.error(new RuntimeException("invalid name!")) cats <- R.get _ <- I.tell(cats.toString) } yield () } 12
  12. Freestyle's Workflow Provide implicit evidence of your handlers to any

    desired target M[_] import cats.effect.IO import cats.effect.implicits._ import cats.data.StateT type Target[A] = StateT[IO, List[String], A] implicit val interactHandler: Interact.Handler[Target] = new Interact.Handler[Target] { def ask(prompt: String): Target[String] = tell(prompt) >> StateT.lift("Isidoro1".pure[IO]) def tell(msg: String): Target[Unit] = StateT.lift(IO { println(msg) }) } implicit val validationHandler: Validation.Handler[Target] = new Validation.Handler[Target] { def minSize(s: String, n: Int): Target[Boolean] = StateT.lift((s.length >= n).pure[IO]) def hasNumber(s: String): Target[Boolean] = StateT.lift(s.exists(c => "0123456789".contains(c)).pure[IO]) } 13
  13. Freestyle's Workflow Run your program to your desired target M[_]

    import modules._ // import modules._ import freestyle.implicits._ // import freestyle.implicits._ import cats.mtl.implicits._ // import cats.mtl.implicits._ val concreteProgram = program[App.Op] // concreteProgram: freestyle.FreeS[modules.App.Op,Unit] = Free(...) concreteProgram.interpret[Target].runEmpty.unsafeRunSync // What's the kitty's name? // List(Isidoro1) // res6: (List[String], Unit) = (List(Isidoro1),()) 14
  14. Effects An alternative to monad transformers error: Signal errors either:

    Flattens if Right / short-circuit Left option: Flatten Some / short-circuit on None reader: Deffer dependency injection until program interpretation writer: Log / Accumulate values state: Pure functional state threaded over the program monadic sequence traverse: Generators over Foldable validation: Accumulate and inspect errors throughout the monadic sequence async: Integrate with callback based API's 15
  15. Effects Error import freestyle.effects.error._ // import freestyle.effects.error._ import freestyle.effects.implicits._ //

    import freestyle.effects.implicits._ type EitherTarget[A] = Either[Throwable, A] // defined type alias EitherTarget def shortCircuit[F[_]: ErrorM] = for { a <- FreeS.pure(1) b <- ErrorM[F].error[Int](new RuntimeException("BOOM")) c <- FreeS.pure(1) } yield a + b + c // warning: there was one feature warning; for details, enable `:setting -feature' or `:replay -feature' // shortCircuit: [F[_]](implicit evidence$1: freestyle.effects.error.ErrorM[F])cats.free.Free[[β$0$]cats.free.FreeApplicative[F,β$0$],Int] shortCircuit[ErrorM.Op].interpret[EitherTarget] // res7: EitherTarget[Int] = Left(java.lang.RuntimeException: BOOM) shortCircuit[ErrorM.Op].interpret[IO].attempt.unsafeRunSync // res8: Either[Throwable,Int] = Left(java.lang.RuntimeException: BOOM) 16
  16. Effects Option import freestyle.effects.option._ // import freestyle.effects.option._ def programNone[F[_]: OptionM]

    = for { a <- FreeS.pure(1) b <- OptionM[F].option[Int](None) c <- FreeS.pure(1) } yield a + b + c // warning: there was one feature warning; for details, enable `:setting -feature' or `:replay -feature' // programNone: [F[_]](implicit evidence$1: freestyle.effects.option.OptionM[F])cats.free.Free[[β$0$]cats.free.FreeApplicative[F,β$0$],Int] programNone[OptionM.Op].interpret[Option] // res9: Option[Int] = None programNone[OptionM.Op].interpret[List] // res10: List[Int] = List() 17
  17. Optimizations with constant evaluation time based on @scala.annotation.switch on the

    Coproduct's internal indexed values. A fast Coproduct type based on Iota import iota._ // import iota._ import iota.debug.options.ShowTrees // import iota.debug.options.ShowTrees val interpreter: FSHandler[App.Op, Target] = CopK.FunctionK.summon // <console>:67: { // class CopKFunctionK$macro$2 extends _root_.iota.internal.FastFunctionK[Op, Target] { // private[this] val arr0 = scala.Predef.implicitly[cats.arrow.FunctionK[algebras.Validation.StackSafe.Op, Target]](algebras.this.Validation. // private[this] val arr1 = scala.Predef.implicitly[cats.arrow.FunctionK[algebras.Interact.Op, Target]](interactHandler).asInstanceOf[_root_. // private[this] val arr2 = scala.Predef.implicitly[cats.arrow.FunctionK[freestyle.effects.error.ErrorM.Op, Target]](freestyle.effects.implic // interpreter: freestyle.FSHandler[modules.App.Op,Target] = FastFunctionK[modules.App.Op, Target]<<generated>> 19
  18. Optimizations Freestyle does not suffer from degrading performance as the

    number of Algebras increases in contrast with cats.data.EitherK 20
  19. Optimizations Optimizations over the pattern matching of FunctionK for user

    defined algebras to translate them into a JVM switch with @scala.annotation.switch. 21
  20. Optimizations Brings ADT-less stack safety to @tagless Algebras without rewriting

    interpreters to Free[M, ?] where M[_] is stack unsafe. program[Option] // Stack-unsafe program[StackSafe[Option]#F] // lift handlers automatically to Free[Option, ?] without the `@free` ADTs overhead 22
  21. Integrations Monix: Target runtime and async effect integration. Fetch: Algebra

    to run fetch instances + Auto syntax Fetch FS. FS2: Embed FS2 Stream in Freestyle programs. Doobie: Embed ConnectionIO programs into Freestyle. Slick: Embed DBIO programs into Freestyle. Akka Http: EntityMarshallers to return Freestyle programs in Akka-Http endpoints. Play: Implicit conversions to return Freestyle programs in Play Actions. Twitter Util: Capture instances for Twitter's Future & Try. Finch: Mapper instances to return Freestyle programs in Finch endpoints. Http4s: EntityEncoder instance to return Freestyle programs in Http4S endpoints.
  22. Standalone libraries (WIP) frees-kafka: Consumer, Producer and Streaming algebras for

    Kafka frees-cassandra: Algebras for Cassandra API's, object mapper and type safe query compile time validation. frees-rpc: Purely functional RPC Services. frees-microservices: Purely functional monitored microservices. 24
  23. Freestyle RPC Define your proto messages import freestyle._ import freestyle.rpc.protocol._

    trait ProtoMessages { @message case class Point(latitude: Int, longitude: Int) @message case class Rectangle(lo: Point, hi: Point) @message case class Feature(name: String, location: Point) @message case class FeatureDatabase(feature: List[Feature]) @message case class RouteNote(location: Point, message: String) @message case class RouteSummary(point_count: Int, feature_count: Int, distance: Int, elapsed_time: Int) } 25
  24. Freestyle RPC Expose Algebras as RPC services import monix.reactive.Observable @option(name

    = "java_package", value = "routeguide", quote = true) @option(name = "java_multiple_files", value = "true", quote = false) @option(name = "java_outer_classname", value = "RouteGuide", quote = true) object protocols extends ProtoMessages { @free @service trait RouteGuideService { @rpc def getFeature(point: Point): FS[Feature] @rpc @stream[ResponseStreaming.type] def listFeatures(rectangle: Rectangle): FS[Observable[Feature]] @rpc @stream[RequestStreaming.type] def recordRoute(points: Observable[Point]): FS[RouteSummary] @rpc @stream[BidirectionalStreaming.type] def routeChat(routeNotes: Observable[RouteNote]): FS[Observable[RouteNote]] } }
  25. Freestyle RPC gives you for free: gRPC Server: gRPC based

    server. client gRPC client. .proto files to interoperate with other langs. 27 Í
  26. Scala FP First approach to .proto generation Find a complete

    example at https: github.com/frees-io/freestyle-rpc-examples syntax = "proto3"; option java_package = "routeguide"; option java_multiple_files = true; option java_outer_classname = "RouteGuide"; message Point { int32 latitude = 1; int32 longitude = 2; } ... service RouteGuideService { rpc getFeature (Point) returns (Feature) {} rpc listFeatures (Rectangle) returns (stream Feature) {} rpc recordRoute (stream Point) returns (RouteSummary) {} rpc routeChat (stream RouteNote) returns (stream RouteNote) {} } 29
  27. Freestyle Microservices Provides a reference impl over RPC optionally including

    Kafka & Cassandra Algebras and Handlers (WIP unfinished design) @free @service trait MyService { @subscribe[Topic.type] def listen(r: ConsumerRecord[Topic#Key, Topic#Value]): FS[Ack] } implicit def myServiceHandler (implicit producer: Producer[Topic#Key, Topic#Value], persistence: Persistence[MyModel]): RouteGuideService.Handler[IO] = new RouteGuideService.Handler[IO] { def listen(r: ConsumerRecord[Topic#Key, Topic#Value]): IO[Ack] = ??? } 30
  28. Brought to you by [colin-passiv](https://github.com/colin-passiv) Adrián Ramírez Fornell <[AdrianRaFo](https://github.com/AdrianRaFo)> Alejandro

    Gómez <[dialelo](https://github.com/dialelo)> Ana Mª Marquez <[anamariamv](https://github.com/anamariamv)> Andy Scott <[andyscott](https://github.com/andyscott)> Diego Esteban Alonso Blas <[diesalbla](https://github.com/diesalbla)> Domingo Valera <[dominv](https://github.com/dominv)> Fede Fernández <[fedefernandez](https://github.com/fedefernandez)> Francisco Diaz <[franciscodr](https://github.com/franciscodr)> Giovanni Ruggiero <[gruggiero](https://github.com/gruggiero)> Javi Pacheco <[javipacheco](https://github.com/javipacheco)> Javier de Silóniz Sandino <[jdesiloniz](https://github.com/jdesiloniz)> Jisoo Park <[guersam](https://github.com/guersam)> Jorge Galindo <[jorgegalindocruces](https://github.com/jorgegalindocruces)> Juan Pedro Moreno <[juanpedromoreno](https://github.com/juanpedromoreno)> Juan Ramón González <[jrgonzalezg](https://github.com/jrgonzalezg)> Maureen Elsberry <[MaureenElsberry](https://github.com/MaureenElsberry)> Peter Neyens <[peterneyens](https://github.com/peterneyens)> Raúl Raja Martínez <[raulraja](https://github.com/raulraja)> Sam Halliday <[fommil](https://github.com/fommil)> Suhas Gaddam <[suhasgaddam](https://github.com/suhasgaddam)> ... and many more contributors 33