how does work scalaz.Validation loopSuccess and loopFailure

Question

Could someone explain with real world examples how does work below methods of scalaz.Validation? I mean loopSuccess and loopFailure.

Snippetes from source code (scalaz7):

scalaz.Validation:

/** Spin in tail-position on the success value of this validation. */
def loopSuccess[EE >: E, AA >: A, X](success: AA => X \/ Validation[EE, AA], failure: EE => X): X =
Validation.loopSuccess(this, success, failure)

/** Spin in tail-position on the failure value of this validation. */
def loopFailure[EE >: E, AA >: A, X](success: AA => X, failure: EE => X \/ Validation[EE, AA]): X =
Validation.loopFailure(this, success, failure)

Companion object:

object Validation extends ValidationFunctions with ValidationInstances {

  /** Spin in tail-position on the success value of the given validation. */
  @annotation.tailrec
  final def loopSuccess[E, A, X](d: Validation[E, A], success: A => X \/ Validation[E, A], failure: E => X): X =
    d match {
      case Failure(e) => failure(e)
      case Success(a) => success(a) match {
        case -\/(x) => x
        case \/-(q) => loopSuccess(q, success, failure)
      }
    }

  /** Spin in tail-position on the failure value of the given validation. */
  @annotation.tailrec
  final def loopFailure[E, A, X](d: Validation[E, A], success: A => X, failure: E => X \/ Validation[E, A]): X =
    d match {
      case Failure(e) => failure(e) match {
        case -\/(x) => x
        case \/-(q) => loopFailure(q, success, failure)
      }
      case Success(a) => success(a)
    }

}

Monroe

Answer 1

This is similar to a trampoline. for loopSuccess, you provide an initial value, and a function which takes you to the next state. There are 3 possible next states:

X.left           // stop processing with X as the result
Success(a).right // no result, run the next iteration with this value
Failure(e).right // stop processing, run the failure function on this result and return it

For loopFailure does the same thing with Failure and Success reversed, so you keep running until you return a left or a Success.

Here is a example of using loopSuccess:

import scalaz._
import Scalaz._

object TestLoopSuccess extends App {
  // check if a number divides another, returning a Failure for division by zero
  val divides : Int => Int => Validation[String,Boolean] = { div => num =>
    if(div == 0) "division by zero".failure
    else (num % div == 0).success
  }

  val allDivide : Int => List[Int] => String \/ Validation[Int,List[Int]] = { div => nums => 
    nums match {
      // empty list means we are done, so we return a left
      case Nil => "All numbers divide".left

      // process the head of the list and return a right
      case x::xs => divides(div)(x).flatMap { divides => 
        if(divides) 
          // head divides, so process more of the list
          xs.success 
        else 
          // head does not divide, so we are done
          "%d is not a multiple of %d".format(x,div).failure
      }.right
    }
  }

  println(Validation.loopSuccess(List(2,4,6,8).success[String], allDivide(0), identity[String])) // "division by zero"
  println(Validation.loopSuccess(List(2,4,6,8).success[String], allDivide(2), identity[String])) // "All numbers divide"
  println(Validation.loopSuccess(List(2,4,7,8).success[String], allDivide(2), identity[String])) // "7 is not a multiple of 2"
}

Virginavirginal