Shouldn’t `EitherT` be covariant? Especially in the Left? - scala

I have the following code
def addKitten(kitten: Kitten): EitherT[Future, GenericError, UUID] = {
val futureOfEither = db.run { // returns a Future[Int] with the number of rows written
kittens += kitten
}.map {
case 1 => kitten.uuid.asRight
case _ => GenericError.SpecificError.asLeft
}
EitherT(futureOfEither)
}
Where SpecificError is a subclass of GenericError. For some reason it does not compile complaining that a SpecificError is not a GenericError. Is it right?
I mean, Either[A, B] should be immutable, so why not making it covariant? Am I missing something?

The same issue for XorT and OptionT was raised here. The reply was:
In Scala, variance has both positives and negatives (perhaps that's how they decided on variance notation! :P). These pros/cons have been discussed numerous times in various venues, so I won't go into them now, but in my opinion at the end of the day you kind of have to settle on "to each their own".
I think this "to each their own" perspective implies that you can't force variance on a type constructor. A concrete example is scalaz.Free in the 7.0 series. It forces the S type constructor to be covariant. At the time I often wanted to wrap a Coyoneda in Free. The most recent versions of Cats and Scalaz have the Coyoneda essentially built into the Free, so this particular use might not be as desired now, but the general principle applies. The problem is that Coyoneda is invariant, so you simply couldn't do this (without a mess of #uncheckedVariance)! By making type constructor parameters invariant, you may end up forcing people to be more explicit about types, but I think that it beats the alternative, where you can prevent them from being able to use your type at all.

Ok, I found a workaround.
I still have no idea why the EitherT is not covariant but you have to remember that the Either itself is covariant.
So the trick is to tell the compiler to use an upper bound for the EitherT creation:
EitherT[Future, GenericError, UUID](futureOfEither)
Having more than one error as left works too (because the compiler is forced to find the LUB) but GenericError has to extend Product and Serializable if it’s a trait and SpecificError is a case class (refer to this question as to why)
...
}.map {
case 1 => kitten.uuid.asRight
case 2 => GenericError.SpecificError2.asLeft
case _ => GenericError.SpecificError.asLeft
}
EitherT(futureOfEither)

As another quite neat workaround you could specify abstract types in flatMap[E, T] type parameters, for example:
// Given ADT
trait Err
case class E1() extends Err
case class E2() extends Err
// We could do (pseudo-code)
EitherT(E1()).flatMap[Err, Int] { x => 100 }
It is in case of FlatMap. For Map you could only transform value and type on the right side.

Related

Scala: Cats, OptionT[Future, T] and ApplicativeError

Some time ago I've started using Cats and found OptionT very useful to work with Future[Option[T]] in most cases. But I faced with one drawback, to use AplicativeError I need to define type alias type FutureOption[T] = OptionT[Future, X] to matching F[_] required by AplicativeError and explicitly specify the type of my expression as FutureOption[T].
type FutureOption[T] = OptionT[Future, T] // definition to match F[_] kind
val x = OptionT.liftF(Future.failed(new Exception("error"))) : FutureOption[String] // need to specify type explicitly
x.recover {
case NonFatal(e) => "fixed"
}
If I remove type definition and explicit type specification of my expression the recover will not be available because OptionT[Future, T] don't match F[_], so it can't be converted implicitly to AplicativeErrorOps.
Unfortunately, the example below won't work because there is no recover method.
val x = OptionT.liftF(Future.failed(new Exception("error")))
x.recover {
case NonFatal(e) => "fixed"
}
Is there any way to avoid such kind of boilerplate code? At least I want to avoid specifying expression types as FutureOption[T] explicitly.
In addition to the other answer, I would like to suggest that you make sure you have -Ypartial-unification enabled for your build.
This is a fix for partial unification of type constructors. You can find a more detailed explanation about the fix here.
With partial unification enabled the code you provided in your question compiles fine. Please note that if you're using an IDE (e.g. Intellij) you might get "false negatives" (the code is underlined as incorrect and code completion doesn't work), but the scalac/sbt/gradle will compile it just fine.
Yes, there are at least two ways to cope with the type ascription.
using type lambdas (this can be intimidating):
val a: { type λ[A] = OptionT[Future, A] }#λ
using a compiler plugin like kind-projector, example usage:
val a: Lambda[A => OptionT[Future, A]]
But if you wanted to call Future's recover, you can always do:
val x = OptionT.liftF(Future.failed(new Exception("error")))
x.value.recover ...

Scala - pattern matching traits that have "self type" as type argument gives unchecked warning

I'm trying to define "immutable setter traits", and generic functions for those.
I have a working implementation, but i'm bit disturbed about the "unchecked" warnings from the pattern matching. I'm not really sure what i can do about it.
type Point = (Double, Double)
trait Sizable[A] {
this: A =>
def size: Point
/* immutable object value setter, returns a new of the same object*/
def size(point: Point): A with Sizable[A]
}
def partitionToSizable[T](elements: List[T]):
(List[T], List[T with Sizable[T]]) =
elements.foldLeft((List[T](), List[T with Sizable[T]]()))((p, c) =>
c match {
case a: T with Sizable[T] => (p._1, p._2 ++ List(a))
case a => (p._1 ++ List(a), p._2)
})
The code above demonstrates the problem.
I'm not even sure how big of an issue is that T being unchecked, since all the elements in the list will have a type of T, and the point of the pattern matching is not to determine if it's type is T since we already know that.
In theory Sizable will always have type of T because of the signature of it's enclosing function.
If it's there is no other solution i'd at least like to suppress the warning. #unchecked annotations does not seem to suppress the warning.
If i modify case a: T with Sizable[T] to case a: Sizable[_] it will not compile, since the result type will obviously not confirm to T.
ClassTags or TypeTags might solve the warning, but i suspect they are not necessary really. (also that might have a performance overhead and TypeTags don't work with Scala.js)
I think just case a: Sizable[T #unchecked] => a.size((15,20)) should work.
The main problem here is in partitionToSizable. It's taking a List[T] and then trying to partition it into a List[T] and a List[T with Sizable[T]]. You are taking a list of a single type and then saying that one type T is actually two different types T and T with Sizable[T]. This indicates an issue with your type design. I would recommend solving that first.
One solution might be to recognise that things that are sizable and things that are not sizable should be represented as two different types. Then you can use the following signature:
def partitionToSizable[T1, T2 <: Sizable[T2]](
elements: List[Either[T1, T2]]): (List[T1], List[T2])
Then you don't need to cast; you can pattern match or fold on the Either.

Implicits over type inference for object transformation

Part of a current project involves converting from types coupled to a database and a generic type used when serializing the results out to clients via Json, the current implementation in Scala uses type inference to correctly perform the transformation, using Scala's TypeTag:
def Transform[A: TypeTag](objects:Seq[A]):Seq[Children] = typeOf[A] match {
case pc if pc =:= typeOf[ProductCategory] =>
TransformProductCategory(objects.asInstanceOf[Seq[ProductCategory]])
case pa if pa =:= typeOf[ProductArea] =>
TransformProductArea(objects.asInstanceOf[Seq[ProductArea]])
case pg if pg =:= typeOf[ProductGroup] =>
TransformProductGroup(objects.asInstanceOf[Seq[ProductGroup]])
case psg if psg =:= typeOf[ProductSubGroup] =>
TransformProductSubGroup(objects.asInstanceOf[Seq[ProductSubGroup]])
case _ =>
throw new IllegalArgumentException("Invalid transformation")
}
The types used as input are all case classes and are defined internally within the application, for example:
case class ProductCategory(id: Long, name: String,
thumbnail: Option[String],
image:Option[String],
sequence:Int)
This approach, although suitable at the moment, doesn't feel functional or scalable when potentially more DB types are added. I also feel using asInstanceOf should be redundant as the type has already been asserted. My limited knowledge of implicits suggests they could be used instead to perform the transformation, and remove the need for the above Transform[A: TypeTag](objects:Seq[A]):Seq[Children] method altogether. Or maybe there is a different approach I should have used instead?
You can define a trait like this:
trait Transformer[A] {
def transformImpl(x: Seq[A]): Seq[Children]
}
Then you can define some instances:
object Transformer {
implicit val forProduct = new Transformer[ProductCategory] {
def transformImpl(x: Seq[ProductCategory]) = ...
}
...
}
And then finally:
def transform[A: Transformer](objects:Seq[A]): Seq[Children] =
implicitly[Transformer[A]].transformImpl(objects)
Preferably, you should define your implicit instances either in Transformer object or in objects corresponding to your category classes.
I'm not sure how your program is supposed to work exactly, nor do I know if any of your Transforms are self-made types or something you pulled from a library, however I may have a solution anyway
Something match is really good with, is case classes
So rather than manually checking the type of the input data, you could maybe wrap them all in case classes (if you need to bring data with you) or case objects (if you don't)
That way you can do something like this:
// this code assumes ProductCategory, ProductArea, etc.
// all extends the trait ProductType
def Transform(pType: ProductType): Seq[Children] = pType match {
case ProductCategory(objects) => TransformProductCategory(objects)
case ProductArea(objects) => TransformProductArea(objects)
case ProductGroup(objects) => TransformProductGroup(objects)
case ProductSubGroup(objects) => TransformProductSubGroup(objects)
}
By wrapping everything in a case class, you can specify exactly which type of data you want to bring along, and so long as they (the case classes, not the data) all inherit from the same class/trait, you should be fine
And since there's only a little handful of classes that extend ProductType you don't need the default case, because there is no default case!
Another benefit of this, is that it's infinitely expandable; just add more cases and case classes!
Note that this solution requires you to refactor your code a LOT, so bare that in mind before you throw yourself into it.

PartialFunction That Isn't Partial

Is there a reason to use a PartialFunction on a function that's not partial?
scala> val foo: PartialFunction[Int, Int] = {
| case x => x * 2
| }
foo: PartialFunction[Int,Int] = <function1>
foo is defined as a PartialFunction, but of course the case x will catch all input.
Is this simply bad code as the PartialFunction type indicates to the programmer that the function is undefined for certain inputs?
There is no advantage in using a PartialFunction instead of a Function, but if you have to pass a PartialFunction, then you have to pass a PartialFunction.
Note that, because of the inheritance between these two, overloading a method to accept both results in something difficult to use, as the type inference won't work.
The thing is, there are many examples of times when what you need to define on a trait/object/function definition is a PartialFunction but in reality the real implementation may not be one. Case in point, take a look at def collect[B](f: PartialFunction[A,B]):
val myList = thatList collect {
case Right(value) => value
case Left(other) => other.toInt
}
It's clearly not a "real" partial as it is defined for all input. That said, if I wanted to, I could just have the Right match.
However, if I were to have written collect as a full on plain function, then I'd miss out on the desired behavior (that is to be both a filter and a map rolled into one base on when a function is defined.) That's nice behavior and allows for a lot of flexibility when writing my own code.
So I guess the better question is, will you ever want behavior to reflect that a function might not be defined everywhere? If the answer is no, then don't do it.
PartialFunction literals allow pattern matching directly on arguments (e.g. { case (a, b) => ... } instead of _ match { case (a, b) => ... }), which makes code more readable (see #wheaties' answer for another example).
EDIT: apparently this is wrong, see Daniel C. Sobral's comment on his answer. Not deleting, so that the comments still make sense.

Map from Class[T] to T without casting

I want to map from class tokens to instances along the lines of the following code:
trait Instances {
def put[T](key: Class[T], value: T)
def get[T](key: Class[T]): T
}
Can this be done without having to resolve to casts in the get method?
Update:
How could this be done for the more general case with some Foo[T] instead of Class[T]?
You can try retrieving the object from your map as an Any, then using your Class[T] to “cast reflectively”:
trait Instances {
private val map = collection.mutable.Map[Class[_], Any]()
def put[T](key: Class[T], value: T) { map += (key -> value) }
def get[T](key: Class[T]): T = key.cast(map(key))
}
With help of a friend of mine, we defined the map with keys as Manifest instead of Class which gives a better api when calling.
I didnt get your updated question about "general case with some Foo[T] instead of Class[T]". But this should work for the cases you specified.
object Instances {
private val map = collection.mutable.Map[Manifest[_], Any]()
def put[T: Manifest](value: T) = map += manifest[T] -> value
def get[T: Manifest]: T = map(manifest[T]).asInstanceOf[T]
def main (args: Array[String] ) {
put(1)
put("2")
println(get[Int])
println(get[String])
}
}
If you want to do this without any casting (even within get) then you will need to write a heterogeneous map. For reasons that should be obvious, this is tricky. :-) The easiest way would probably be to use a HList-like structure and build a find function. However, that's not trivial since you need to define some way of checking type equality for two arbitrary types.
I attempted to get a little tricky with tuples and existential types. However, Scala doesn't provide a unification mechanism (pattern matching doesn't work). Also, subtyping ties the whole thing in knots and basically eliminates any sort of safety it might have provided:
val xs: List[(Class[A], A) forSome { type A }] = List(
classOf[String] -> "foo", classOf[Int] -> 42)
val search = classOf[String]
val finalResult = xs collect { case (`search`, result) => result } headOption
In this example, finalResult will be of type Any. This is actually rightly so, since subtyping means that we don't really know anything about A. It's not why the compiler is choosing that type, but it is a correct choice. Take for example:
val xs: List[(Class[A], A) forSome { type A }] = List(classOf[Boolean] -> 'bippy)
This is totally legal! Subtyping means that A in this case will be chosen as Any. It's hardly what we want, but it is what you will get. Thus, in order to express this constraint without tracking all of the types individual (using a HMap), Scala would need to be able to express the constraint that a type is a specific type and nothing else. Unfortunately, Scala does not have this ability, and so we're basically stuck on the generic constraint front.
Update Actually, it's not legal. Just tried it and the compiler kicked it out. I think that only worked because Class is invariant in its type parameter. So, if Foo is a definite type that is invariant, you should be safe from this case. It still doesn't solve the unification problem, but at least it's sound. Unfortunately, type constructors are assumed to be in a magical super-position between co-, contra- and invariance, so if it's truly an arbitrary type Foo of kind * => *, then you're still sunk on the existential front.
In summary: it should be possible, but only if you fully encode Instances as a HMap. Personally, I would just cast inside get. Much simpler!