WTF since when does the following result in an amb...

# random

[robstoll]

WTF since when does the following result in an ambiguity?

class A<T>{}
fun <E> A<E>.foo(a: E){}
fun <E> A<E>.foo(b: Int){}
fun test(){
    A<Int?>().foo(null)
}

[Nikky]

T and E are not the same though right? so just because A<Int?> the function `foo`'s generic is not known

[Nikky]

although it should definitly not match the second function, so i do not know whats up with that

[robstoll]

it does not resolve null to Int? for whatever reason

[robstoll]

the following compiles:

class A<T>{}
fun <E: Any?> A<E>.foo(a: E){}
fun <E: Any> A<E>.foo(b: E){}
fun test2(){
    A<Int?>().foo(null)
}

[robstoll]

I'll file an issue as soon as youtrack is not in read only mode

[Nikky]

i think whats happening is that it means that this will result in a clash in bytecode

[Nikky]

Int or Int? are the same there

[robstoll]

but overload resolution is done at compile time not at runtime

[Dominaezzz]

Wait, isn't that ambiguous?

[Dominaezzz]

Oh, I just understood lol.

[robstoll]

also, there is no byte code clash (at least should not)

[robstoll]

you can write the following in java as well:

public class AExtension {
    public static <E> void  foo(E e){}
    public static void  foo(Integer e){}
}

[Dominaezzz]

Shouldn't that be int?

[robstoll]

true 😄

[robstoll]

but does not really matter for the sake of illustration 😉

[Dominaezzz]

Yeah, it's probably some edge case with "primitives" in the Kotlin compiler.

[robstoll]

replace Int with String the problem remains

[robstoll]

https://youtrack.jetbrains.com/issue/KT-6591

[robstoll]

quite an old bug as it seems

[Dominaezzz]

4 years, I don't think they're gonna fix it at this point.

[robstoll]

seems to be fixed in the new type inference system

[robstoll]

so yeah... we need to wait until it is released I guess

[Dominaezzz]

(sigh). Why do I keep being wrong.

[robstoll]

Shouldn't that be int?

That was correct 😉