Hi! Why is type inference failing here? To my eyes it has all the info it needs to infer the type...

class Builder {
  fun <R> build(action: (R) -> R): Builder = this
}

fun main() {
  val value = 3
  Builder()
    .build { value } // compiler error: not enough information to infer type variable R
}

That is: lambda is returning

Int

therefore

R

is

Int

UPD: this is kotlin 1.5.0

Because “value” is not explicitly “Int”, it can be at least a Double or Float as well

no, it's

Int

. that's kotlin's default, it has no ambiguity in this case. If I add

val value: Int = 3

it highlights

: Int

as "redundant" and the error still stays. Even if I replace int with string, or whatever other type, it stays, so that's not it.

I’m guessing it’s because there’s not enough info to infer the input parameter of the lambda. It seems to work if you change

action

to be

() -> R

or if you do

Builder().build { _: Int -> value }

Indeed! But I wonder why is it so? I've reduced it to a simple function:

fun <R> build(action: (R) -> R) = Unit

fun main() {
  val value: Int = 3
  build { value } // not enough information to infer type variable R
}

But here's an equivalent Haskell program and it compiles and runs perfectly. So it seems like a type inference limitation...

build :: (a -> a) -> ()
build action = ()

main = do
  let value = 3
  return $ build (\_ -> value)

Kotlin has more type inference than Java but not as much as a full H-M system like Haskell (none of which work freely with polymorphism, so don't expect that)

It seems

.build<Int> { value }

looks cleaner to me. There is

BuilderInference

, but I don't know if it can be applied here; afaik it infers type from value arguments.

for this case, yes.

.build { _: Int -> value }

is sometimes preferable if there are multiple type parameters and you only need to hint a single one (because

<>

forces you to write all of them)

I wonder can this be done at all in kotlin compiler, maybe I should file a feature request.

I think you're asking for https://youtrack.jetbrains.com/issue/KT-8138 although as a QoL improvement I'd like https://youtrack.jetbrains.com/issue/KT-13394 too

Thanks, subscribed! I'll add my case to the second one. However. No, it doesn't look exactly like the second one. Will try to report a new one if I won't find existing issue.