I have an app that does a lot of calls to different backend systems, and hoping to use for-comprehensions to simplify the process flow across the backend systems.
I'm looking to combine EitherT (scalaz) and Future (scala 2.10) so I can capture the first potential error (where its a future or backend system issue) and return an appropriate message to the end user. I've had a quick look a scalaz Validation but the recommendation for capturing first error and not all errors is to use EitherT.
I'm trying a simple example in REPL first, however I'm getting the following error
error: could not find implicit value for parameter F: scalaz.Functor[scala.concurrent.Future]
import scala.concurrent._
import scalaz._
import Scalaz._
import ExecutionContext.Implicits.global
type EitherFuture[+A] = EitherT[Future, String, A]
def method1Success : EitherFuture[Int] = {
println("method 1 success")
EitherT {
Future {
1.right
}
}
}
def method2Failure : EitherFuture[Int] = {
println("method 2 failure")
EitherT {
Future {
"fail".left
}
}
}
val m1 = method1Success
// problem
m1.isRight
// problem
def methodChain1 = {
for {
a <- method1Success
b <- method2Failure
} yield b
}
I'm still new to both scala and scalaz so any pointers would be great.
** Update **
By including scalaz-contrib based on @stew suggestion I now have an updated version that shows for-comprehensions with the combined EitherT and Future showing different simple use cases backend success, a backend failure, and a future failure
import scala.concurrent._
import scalaz._
import Scalaz._
import ExecutionContext.Implicits.global
import scalaz.contrib._
import scalaz.contrib.std._
import scala.concurrent.duration._
type EitherFuture[+A] = EitherT[Future, String, A]
// various methods that mimic success or different failures
def methodBackendSuccess : EitherFuture[Int] = {
println("method backend success")
EitherT {
Future {1.right}
}
}
def methodBackendFailure : EitherFuture[Int] = {
println("method backend failure")
EitherT {
Future { "fail".left}
}
}
def methodFutureFailure : EitherFuture[Int] = {
println("method future failure")
EitherT {
Future.failed(new Exception("future failed"))
}
}
// different combinations for for-comprehensions
def methodChainBackendSuccess = {
for {
a <- methodBackendSuccess
b <- methodBackendSuccess
c <- methodBackendSuccess
} yield c
}
def methodChainBackendFailure = {
for {
a <- methodBackendSuccess
b <- methodBackendFailure
c <- methodBackendSuccess
} yield c
}
def methodChainFutureFailure = {
for {
a <- methodBackendSuccess
b <- methodFutureFailure
c <- methodBackendSuccess
} yield c
}
// process results for different chain methods
def processOutcome(chainMethod: => EitherFuture[Int]):Int = try {
val x = Await.result(chainMethod.run, 30 seconds)
x.toEither match {
case Left(l) => {
println("Backend failure <" + l + ">")
-1
}
case Right(r) => {
println("Backend success <" + r + ">")
r
}
}
} catch {
case e: Exception => {
println("Future error <" + e.getMessage + ">" )
-99
}
}
// run tests
val backendSuccess = processOutcome(methodChainBackendSuccess)
val backendFailure = processOutcome(methodChainBackendFailure)
val futureFailure = processOutcome(methodChainFutureFailure)
Either
business at all?Future
can already model failure, and sequencing has the behavior you want. – Radioactivate"org.typelevel" %% "scalaz-contrib-210" % "0.2"
it should work. Not the double%%
and removal of_2.11
– Soundboard