I have a weird construction for which IntelliJ com...

# intellij

[Davio]

I have a weird construction for which IntelliJ complains:

class Test {
    suspend fun test1() = test3<String>(::test2)
    suspend fun test2() = withContext(<http://Dispatchers.IO|Dispatchers.IO>) { }

    private suspend inline fun <reified T> test3(block: suspend () -> Unit): List<T> {
        block.invoke()
        return listOf()
    }
}

IntelliJ suggests that I can remove the suspend keyword from the lambda parameter of test3, but if I remove it, it no longer compiles.
The reason IntelliJ gives is Redundant 'suspend' modifier: lambda parameters of suspend function type uses existing continuation.

Is this an error in IntelliJ or the compiler?

[Davio]

This is a simplification of my actual code, in my actual code, I need access to T's class inside the private function

[Sam]

The compile error after you remove the suspend modifiers from the inline function just comes down to the type of the function reference ::test2. I guess the compiler doesn’t have a rule to say that in the specific context of an inlined function parameter, it’s okay to convert from suspend () -> Unit to () -> Unit.

[Sam]

You can work around it by writing test3<String> { test2() } instead of test3<String>(::test2). Since it’s an inlined lambda, the result should be identical.

[Starr]

I think there might be a known issue regarding suspend inline method references

[Sam]

Might be this one: https://youtrack.jetbrains.com/issue/KT-30703/Support-suspend-inline-function-conversions-for-callable-reference