Strange think in stdlib-common I found, looks like a bug. Details in comments. (`setOf(singleElement...

# multiplatform

[Tomasz Krakowiak]

Strange think in stdlib-common I found, looks like a bug. Details in comments. (setOf(singleElement))

[Tomasz Krakowiak]

/**
 * Returns a new read-only set with the given elements.
 * Elements of the set are iterated in the order they were specified.
 * The returned set is serializable (JVM).
 * @sample samples.collections.Collections.Sets.readOnlySet
 */
public fun <T> setOf(vararg elements: T): Set<T> = if (elements.size > 0) elements.toSet() else emptySet()

/**
 * Returns a [Set] of all elements.
 * 
 * The returned set preserves the element iteration order of the original array.
 */
public fun <T> Array<out T>.toSet(): Set<T> {
    return when (size) {
        0 -> emptySet()
        1 -> setOf(this[0])
        else -> toCollection(LinkedHashSet<T>(mapCapacity(size)))
    }
}

How, the heck is it supposed to work for array of size 1? ; p I guess on runtime we never actually use both of those implementation... ?

[Luis]

What is the issue you are seeing? Looks fine to me

[kqr]

cycle?

[Tomasz Krakowiak]

Cycle indeed, infinite recursion.

[Mayank]

There is another function in jvm and native source-set which takes only one element

public fun <T> setOf(element: T): Set<T>

I think this function is being used instead of one with the vararg

[Tomasz Krakowiak]

@Mayank Shouldn't then there be an except declaration in common?

[Mayank]

As per my understanding, many files in the stdlib are generated including the _Arrays.kt and due to method overloading, platform specific setOf(element: T) is used if present otherwise it fall backs to setOf(vararg elements: T)

[Mayank]

as both the methods are in the same package, imports need not be changed