gadthelp

gadthelp.scala

trait AA

trait AA1
trait BB
trait BB1

trait HasScenarios[A, B] {
  def scenarios: Map[String, Scenario[A, B]]
}

sealed trait Scenario[A, B]

object Scenario {
  // later pattern matching let's me do GADT stuff in pattern matches fine
  case object A extends Scenario[AA, AA1]
  case object B extends Scenario[BB, BB1]

  def from[A, B](s: String)(implicit hasScenarios: HasScenarios[A, B]): Scenario[A, B] = hasScenarios
    .scenarios
    .getOrElse(s, ??? /* fail the parser here */ )

}


// later usage using the GADT in a pattern match
object Generator {
  def from[A, B](scenario: Scenario[A, B]): Box[A, B] = scenario match {
    case A     => constantA // returns some Box[AA, AA1]
    case B     => constantB // returns some Box[BB, BB1]
  }
}

// later later usage 
// relies on the answer of Box[A, B] to lookup implicits for A, B. 
  // I can use an existential type to parse the string and return Scenario[_,_] later
  // but then I lose the fact it's AA, AA1 or BB, BB1 later 
  // and a bunch of implicit lookups fail because it can't find implicits for
  // _$1, and _$2 

/*

e.g. I have some 

for { 
 .. some stuff 
 scenario = Scenario.from(someString) // will be Either[Throwable, Scenario[_, _]].liftTo[IO] or whatever later 
 generator = Generator.from(scenario) 
 thing <- doThing(generator)
  ...
  } yield ... 



where doThing is 

def doThing[A : Codec, B: Codec](box: Box[A, B], ...) { 

} 

but if I type the initial parsing as a existential type
I get the issue 
could not find implicit Codec[Box[_$1,_$2]]]; ensure no imports are missing 

but there are definitely codecs that can be derived for Box[AA, AA1] and Box[BB, BB1] 
it just is confused as I've lost the typing, even though I should be getting it back 
on the pattern match in the gadt to go from Scenario -> Box[A, B] 
*/