Sash translates regular Scala code into monadic expressions via a blackbox (meaning that it is mostly transparent to your IDE typechecker) macro. Unlike some alternatives, Sash clearly splits the code it translates into statements, which are chained together via some version of flatMap, and expressions, which are searched for effectful subexpressions. This approach eliminates the need for the infamous _ <- EFFECT construct or its equivalents:

import com.github.mvv.sash.cats._
import cats.effect._
import cats.effect.concurrent.Ref
import cats.effect.Console.io._

def askFor(thing: String): IO[Unit] = effect[IO] {
  def ask = effect {
    try readLn
    catch {
      case e: IOException =>
        putStrLn("Yikes!")
        throw e
    }
  }
  putStrLn(s"Enter $thing")
  val answer = +Ref.of[IO, String](+ask)
  while (+answer.get != thing) {
    putStrLn(s"No, enter $thing")
    answer.set(+ask)
  }
  putStrLn("You pass")
}

or, with ZIO:

import com.github.mvv.sash.zio._
import zio._
import zio.console._
import java.io.IOException

def askFor(thing: String): ZIO[Console, IOException, Unit] = effect[Console, IOException] {
  def ask = effect {
    try getStrLn
    catch {
      case e: IOException =>
        putStrLn("Yikes!")
        throw e
    }
  }
  putStrLn(s"Enter $thing")
  val answer = +Ref.make(+ask)
  while (+answer.get != thing) {
    putStrLn(s"No, enter $thing")
    answer.set(+ask)
  }
  putStrLn("You pass")
}

The only noticeable difference from a regular Scala code is the prolific use of +EFFECT construct, which is roughly equivalent to VAL <- EFFECT in for-comprehensions. For example

val answer = +Ref.make(+ask)

would look like

tmp    <- ask
answer <- Ref.make(tmp)

and

answer.set(+ask)

would look like

tmp <- ask
_   <- answer.set(tmp)

inside a for.

Core Sash module

libraryDependencies += "com.github.mvv.sash" %% "sash" % "0.1-M7"

provides a "simple" version of the effect macro, which relies only on flatMap method. It can handle conditionals and loops, but cannot handle try-catch-finally. The module also exposes the imlementation of the macro, which can be configured to handle your favourite monads via a small compatibility layer. Sash comes with two such layers: one for the Cats library

libraryDependencies += "com.github.mvv.sash" %% "sash-cats" % "0.1-M7"

and one for the ZIO

libraryDependencies += "com.github.mvv.sash" %% "sash-zio" % "0.1-M7"

Translation starts with the argument of the macro, which is treated as a statement. A statement can be

• A unit value ()
• A block { [STMT]* } of statements
• A conditional if (EXPR) STMT [else STMT]
• A match EXPR match { [case ... => STMT]* }
• A loop while (EXPR) STMT or do STMT while (EXPR)
• A variable declaration [implicit] val NAME[: TYPE] = EXPR
• An error raising statement throw EXPR
• An error handling statement try STMT [catch { [case ... => STMT]* }] [finally STMT]
• An import or a type/class/trait/object/function definition. Those are left left as-is, meaning that they are simply brought into scope of the subsequent statements.
• Impure code impure CODE, where CODE is a regular Scala code
• An expression EXPR

Expressions EXPR are analyzed further, to see if they are

• Effectful +STMT
• Pure pure CODE, where CODE is a regular Scala code
• Typed EXPR: TYPE
• An application EXPR([EXPR]*)
• An accessor EXPR.NAME
• A conditional if (EXPR) CODE else CODE
• A match EXPR match { [case ... => CODE] }
• A regular Scala code CODE