Collections and Sequences

Read-only by default

Kotlin distinguishes read-only interfaces (List, Set, Map) from mutable ones (MutableList and so on), created with listOf or mutableListOf. The standard library offers many operations: map, filter, flatMap, associateBy, groupBy, partition, sortedBy, sumOf, fold, reduce, any, all, firstOrNull, chunked, windowed and more. Each collection operation is eager and creates an intermediate list. A sequence (asSequence() or sequence { }) is lazy: elements pass through the whole chain one at a time, and processing stops early with operations like first or take. Sequences help with long chains on large data; for small collections plain lists are usually fine.

Summarising orders

Eager list operations and a lazy sequence.

data class Order(val customer: String, val total: Double, val paid: Boolean)

fun main() {
    val orders = listOf(
        Order("ada", 30.0, true), Order("linus", 12.5, false), Order("ada", 7.5, true)
    )

    val revenueByCustomer: Map<String, Double> = orders
        .filter { it.paid }
        .groupBy { it.customer }
        .mapValues { (_, list) -> list.sumOf { it.total } }

    val (paid, unpaid) = orders.partition { it.paid }
    val count = orders.fold(0) { acc, order -> if (order.total > 10) acc + 1 else acc }

    val firstBigSquare = generateSequence(1) { it + 1 }   // infinite, lazy
        .map { it * it }
        .first { it > 1_000 }
}

Use OrNull variants

Functions like first() and single() throw on no match; firstOrNull(), singleOrNull() and getOrNull(i) return null and combine nicely with ?:.

Quick check: What is the key difference between a List chain and a Sequence chain?

• Sequences are mutable lists
• Sequences are always faster
• Lists cannot be filtered
• Sequences are evaluated lazily, element by element