User Guide¶

kotlin-data-ref is intentionally small. It exists for the cases where Kotlin's default structural collection semantics are wrong for the job.

Identity Versus Equality¶

Ordinary Kotlin collections use equals and hashCode.

data class Node(val id: Int)

val a = Node(1)
val b = Node(1)

check(a == b)
check(setOf(a, b).size == 1)

An identity set keeps both objects because they are distinct references:

import one.wabbit.data.identitySetOf

val identity = identitySetOf(a, b)
check(identity.size == 2)

Identity Maps¶

Use identityMap when the key object itself is the identity:

import one.wabbit.data.identityMap

val values = identityMap<Node, String>()
values[a] = "first"
values[b] = "second"

check(values[a] == "first")
check(values[b] == "second")

This is useful in object graphs, AST traversals, memoization over mutable objects, and cycle detection where structural equality can merge nodes that must stay separate.

The returned value is still a MutableMap. Key lookup, insertion, and removal use identity, while the entries view follows the map-entry contract with identity matching for keys and structural matching for values.

Distinct By Identity¶

distinctByIdentity removes repeated references while preserving the first occurrence:

import one.wabbit.data.distinctByIdentity

val node = Node(1)
val result = listOf(node, Node(1), node).distinctByIdentity()

check(result.size == 2)
check(result.first() === node)

Use the sequence overload when the input is already a sequence and you want lazy filtering.