[SI-2712] implement higher-order unification for type constructor inference Created: 27/Nov/09  Updated: 18/Oct/16  Resolved: 27/May/16

Status: CLOSED
Project: Scala Programming Language
Component/s: Compiler (Misc)
Affects Version/s: None
Fix Version/s: Scala 2.11.9, Scala 2.12.0-M5

Type: Improvement Priority: Minor
Reporter: Adriaan Moors Assignee: Adriaan Moors
Resolution: Fixed Votes: 83
Labels: None


Issue Links:
blocks SI-5993 Unexpected compiler error involving t... CLOSED
is duplicated by SI-6744 It is impossible to pattern match on ... CLOSED
is mentioned by SI-6895 Type class instance can't be found fo... CLOSED
is mentioned by SI-9361 Compiler crash inferring higher-kinde... CLOSED
relates to SI-4719 multiple type parameter lists Open
relates to SI-7745 Extension method interferes with unif... Open
relates to SI-8286 Implicit search fails to expand type ... Open
relates to SI-5075 Anonymous type function is accepted a... Open
Alan, arjan blokzijl, Christos KK Loverdos, Eric Torreborre, Evgeny, Ismael Juma, Ittay Dror, Johannes Rudolph, Jorge Ortiz, Josh Suereth, Kris Nuttycombe, Mark Harrah, Michael, Miles Sabin, Nick Partridge, Seth Tisue, Tony Morris, weihsiu


implement what's described in "Partial polymorphic type inference and higher-order unification"

object Test {
  def meh[M[_], A](x: M[A]): M[A] = x
  meh{(x: Int) => x} // should solve ?M = [X] X => X and ?A = Int ... 

Comment by Harrison Klaperman [ 05/Jun/11 ]

You can find that paper here: http://www.cs.cmu.edu/~fp/papers/lfp88.pdf. I hope this is useful to others who stumble upon this issue. Now if only I understood what the paper were talking about...

Comment by Paolo G. Giarrusso [ 19/Oct/11 ]

Further work on that problem is described in "Unification via explicit substitutions: The case of higher-order patterns" - http://www.cs.cmu.edu/~fp/papers/INRIA-RR-3591.pdf, from Dowek, Hardin, Kirchner and Pfenning (Pfenning is the author of the above paper). I am not an expert, but this work is cited in Pierce's TAPL (Sec. 23.6, pag. 355), which explains that unlike the algorithm above it is terminating and always finds a unique solution; the paper suggests that this work applies to an important subcase. However, I cannot fully judge myself the respective merits of the two papers.

Comment by Miles Sabin [ 06/Nov/11 ]

With dependent method types we have a somewhat cumbersome, but just about usable workaround. The key idea is to use implicits to destructure a type, and then pick out and use its component parts as dependent types.

// Template we will unpack into
trait UnpackM[MA] {
  type M[_]
  type A
// Destructuring implicits
implicit def unpackM1[M0[_], A0] = new UnpackM[M0[A0]] {
  type M[X] = M0[X]
  type A = A0
implicit def unpackM2[M0[_, _], A0, B0] = new UnpackM[M0[A0, B0]] {
  type M[X] = M0[A0, X]
  type A = B0
def meh[MA](x : MA)(implicit u : UnpackM[MA]) : (u.M[String], List[u.A]) = null
//      ^^          ^^^^^^^^^^^^^^^^^^^^^^^^     ^^^^^^^^^^^       ^^^
//      (1)                   (2)                    (3)           (3)
// 1: Type variable being destructured
// 2: Destructuring implicit
// 3: Uses of (dependent) component types
val m = meh{(x: Int) => x}
implicitly[m.type <:< (Int => String, List[Int])]

See also here for a slightly more elaborate scalaz inspired example.

Comment by Paolo G. Giarrusso [ 09/Dec/11 ]

Milles Sabin, your workaround is interesting:
But unfortunately the example you link still needs these two lines:

type EitherInt[A] = Either[Int, A]
implicit object EitherTC extends TC[EitherInt]

Moreover, your code does not even correctly solve the original problem - it only infers that Int => Int matches M0[A0, B0] with M0 = Function1, not that it matches M1[A1] with M1[X] = X => X.

Comment by Jason Zaugg [ 02/Dec/12 ]

See also SI-6744, where we would like to infer these sort of types in pattern matching.

Comment by Paul Chiusano [ 02/Dec/12 ]

Would it be any easier to just look for partial applications of existing type constructors in left-to-right order? Haskell does roughly this (actually, type constructors are just curried, see below), it is tractable, and people don't seem to have an issue with the limitation, though occasionally you do have to introduce a new type just to flip the order of some type parameters.

Here's an example:

case class State[S,A](run: S => (A,S))
object Blah {
  def foo[M[_],A](m: M[A]): M[A] = m
  val blah = foo(State((i: Int) => ("hi",i+1)))

Another way to think about this is that type constructors are curried, so State[S,A] is actually represented as Ap(Ap(State, S), A), and likewise the M[A] is Ap(freevar, A) - these unify with freevar = Ap(State, S).

Comment by Dan Rosen [ 19/Apr/13 ]

To Paul's comment, I agree that looking at the type parameters as curried, partially applied from left to right (SI-4719) is probably usually good enough.

If State[S, A] has pseudo-Haskell-ish kind (*, *) -> *, then State[S][A] would have kind * -> * -> *, and State[S] would have kind * -> *.

Comment by Paolo G. Giarrusso [ 17/Sep/14 ]

It seems somehow I never mentioned higher-order pattern unification by Miller, which is what Agda uses — the paper I linked ("Unification via explicit substitutions: The case of higher-order patterns") is indeed about that. For whatever reason, everybody steers away from the original Huet algorithm, including modern implementation of lambdaProlog (which is what "Partial Polymorphic Type Inference and Higher-Order Unification" uses).

Comment by Adriaan Moors [ 30/Jun/15 ]

Oh, jira.

Comment by Seth Tisue [ 30/Jun/15 ]

SI-2712 closed!! Dewey defeats Truman!!

Comment by Miles Sabin [ 01/Jul/15 ]

That was cruel and heartless ...

Comment by Eric Torreborre [ 01/Jul/15 ]

My heart rate went up for a moment...

Comment by Stephen Compall [ 01/Jul/15 ]

long, wistful sigh

Comment by Jed Wesley-Smith [ 01/Jul/15 ]

well, that was (briefly) exciting!

Comment by Matthew Pocock [ 23/Jul/15 ]

I keep on hitting this issue in various forms, most recently when trying to use type lambdas to fill in one of two types on a * -> * -> * kinded type. I have this type:

case class RangeAs[R, V](range: R, lower: V, upper: V)

object RangeAs {
trait Aux[V]

{ type l[R] = RangeAs[R, V] }


There are various implicit instances with RangeAs.Aux[X]#l as one of their type parameters. They are not found in all situations if I attempt to summon them using the lambda. However, if I do this:

type RangeAsX[T] = RangeAs.Aux[X]#l[T]

Now if I summon them using RangeAsX, they are found.

Comment by Ismael Juma [ 12/Apr/16 ]

For the watchers of this JIRA that haven't seen this on Twitter, Miles posted a demo project exercising a proposed fix for this JIRA:


Comment by Miles Sabin [ 15/Apr/16 ]

PR against 2.12.x here: https://github.com/scala/scala/pull/5102.

Comment by Miles Sabin [ 26/Aug/16 ]

Backport to 2.11.8 available here.

Comment by Adriaan Moors [ 18/Oct/16 ]

2.11.9 PR: https://github.com/scala/scala/pull/5343

Generated at Fri Jul 20 09:02:10 CEST 2018 using JIRA 7.9.1#79001-sha1:60970b42586a2ec2760ed6cfe825b26961e62b9e.