[scala] type checking problem
|
View:
New views
3 Messages
—
Rating Filter:
Alert me
|
 |
[scala] type checking problem
by daniel mahler
::
Rate this Message:
in Scala-2.7.4 & current 2.80 snapshots I get the following:
scala> object Test { | | trait ZZ { | trait C1[T]{} | | trait CF1[T, CT <: C1[T]]{} | | trait A[T] { | type R <: C1[T] | type FR <: CF1[T, R] | def fr : FR | } | | def f1[Y, AY <: A[Y]](y: Y, ay: AY) : AY#R | = f2[Y, AY#R](y, ay.fr : AY#FR) | | def f2[O, CO <: C1[O]](f: O, fr : CF1[O, CO]) : CO | } | } <console>:21: error: type mismatch; found : AY#FR required: ZZ.this.CF1[Y,AY#R] = f2[Y, AY#R](y, ay.fr : AY#FR) ^ but I think AY#FR <: CF1[Y,AY#R] so why is this an error? thanks Daniel |
|
|
[scala] Re: type checking problem
by daniel mahler
::
Rate this Message:
On Sun, Oct 11, 2009 at 11:58 PM, Daniel Mahler <dmahler@...> wrote: in Scala-2.7.4 & current 2.80 snapshots I get the following: I have found that I this code will compile if I make CF1 covariant in CT. object Test { trait ZZ { type C1[T] type CF1[T, +CT <: C1[T]] trait A[T] { type R <: C1[T] type FR <: CF1[T, R] def fr : FR } def f1[Y, AY <: A[Y]](y: Y, ay: AY) : AY#R = f2[Y, AY#R](y, ay.fr) def f2[O, CO <: C1[O]](f: O, fr : CF1[O, CO]) : CO } } Unfortunately in my intended application CF1 is not covariant in CT and I am havingtrouble figuring out what is forcing the covariance requirement. Is there some other minor change that would make the above work? The f2, C1 & CF1 reperesent fixed third party library interfaces; though C1 & CF1 could be covariant in T. I am trying to package up the fr, FR & R components and provide a cleaner interface to f2 via f1 and ultimately make the ay parameter implicit (the reason for trying to package things together in the first place) If I remove the covariance and look at the -explaintypes output there is one line that puzzles me: experimental/users/mahler/java/porn/Tests.scala:19: error: type mismatch; found : ay.FR required: ZZ.this.CF1[Y,AY#R] = f2[Y, AY#R](y, ay.fr) ^ ay.FR < ZZ.this.CF1[Y,AY#R]? ay.FR < Nothing? <notype> < Nothing? false ZZ.this.CF1[Y,ay.R] < Nothing? <notype> < Nothing? false Any < Nothing? <notype> < Nothing? false false false false ZZ.this.CF1[Y,ay.R] < ZZ.this.CF1[Y,AY#R]? AY#R < ay.R? *** Why does the ordering flip in the previous step? AY < ay.type? ZZ.this.A[Y] < ay.type? false false AY#R < Nothing? <notype> < Nothing? false ZZ.this.C1[Y] < Nothing? <notype> < Nothing? false Any < Nothing? <notype> < Nothing? false false false false ZZ.this.C1[Y] < ay.R? ZZ.this.C1[Y] < Nothing? <notype> < Nothing? false Any < Nothing? <notype> < Nothing? false false false Any < ay.R? Any < Nothing? <notype> < Nothing? false false false false false ZZ.this.CF1[Y,ay.R] < Nothing? <notype> < Nothing? false Any < Nothing? <notype> < Nothing? false false false Any < ZZ.this.CF1[Y,AY#R]? Any < Nothing? <notype> < Nothing? false false false false false [[syntax trees at end of typer]]// Scala source: Tests.scala package <empty> { final object Test extends java.lang.Object with ScalaObject { def this(): object Test = { Test.super.this(); () }; abstract trait ZZ extends java.lang.Object with ScalaObject { def /*ZZ*/$init$(): Unit = { () }; type C1[T >: Nothing <: Any]>: Nothing <: Any; type CF1[T >: Nothing <: Any, CT >: Nothing <: ZZ.this.C1[T]]>: Nothing <: Any; abstract trait A[T >: Nothing <: Any] extends scala.AnyRef { type R>: Nothing <: ZZ.this.C1[T]; type FR>: Nothing <: ZZ.this.CF1[T,A.this.R]; def fr: A.this.FR }; def f1[Y >: Nothing <: Any, AY >: Nothing <: ZZ.this.A[Y]](y: Y, ay: AY): AY#R = ZZ.this.f2[Y, AY#R](y, ay.<error: method fr>); def f2[O >: Nothing <: Any, CO >: Nothing <: ZZ.this.C1[O]](f: O, fr: ZZ.this.CF1[O,CO]): CO } } } Thanks Daniel |
|
|
[scala] Re: type checking problem
by daniel mahler
::
Rate this Message:
I figured out my problem.
What I was actually after is: object Test { trait ZZ { type C[O] type CF[O, CO <: C[O]] trait A[O, CO <: C[O]] { type CFO <: CF[O, CO] def cfo : CFO } def f1[O, CO <: C[O]](o: O, ao: A[O, CO]) : CO = f2[O, CO](o, ao.cfo) def f2[O, CO <: C[O]](o: O, cfo : CF[O, CO]) : CO } Though, I suspect the benefit of these contortions is not apparent without the context :) cheers Daniel On Mon, Oct 12, 2009 at 3:46 AM, Daniel Mahler <dmahler@...> wrote:
|
| Free embeddable forum powered by Nabble | Forum Help |
