shouldn't `inline fun <reified T> foo() = T::class.qualifiedName` be disallowed if `inline fun...

# announcements

[robstoll]

shouldn't inline fun <reified T> foo() = T::class.qualifiedName be disallowed if inline fun <reified T> foo() = T::class is?

[Shawn]

It’s not disallowed per se, the generics just make things a bit more complicated

[Shawn]

inline fun <reified T : Any> foo(): KClass<T> = T::class

is what you want because KClass<T> is explicitly bounded to Any

[robstoll]

exactly, so why is this ignored in inline fun <reified T> foo() = T::class.qualifiedName or to ask differently what is the type of T::class and how can I use it in other code 😄

[Shawn]

because the return type is inferred as String?

[Shawn]

and you don’t need to infer a projection or bound for T in the return type

[Shawn]

the type of T::class is KClass<T>

[robstoll]

I don't think so, as KClass<T> requires T to have an upper bound of Any -- it's unsafe to ignore that no?

[robstoll]

inline fun <reified T> foo() : String? {
    val a: KClass<T> = T::class
    return a.qualifiedName
}

Does not compile

[robstoll]

btw. foo<Int?>() returns <http://kotlin.Int|kotlin.Int>, no mentioning of the nullable part

[Shawn]

I noticed that, unsure how that resolves exactly

[Shawn]

especially since Int?::class won’t compile

[robstoll]

it looks a bit like a bug no?

[Shawn]

honestly I don’t know enough to say