Operations with streams: interview questions & scenarios

Model answer

map is a one-to-one transform. If you start with Stream<Order> and call .map(Order::id), each order becomes exactly one id, so the result is Stream<String>.

flatMap is zero-to-many or one-to-many. If each order has a list of line items, .map(Order::lines) gives a Stream<List<LineItem>>, while .flatMap(order -> order.lines().stream()) gives one flattened Stream<LineItem>.

List<String> skus = orders.stream()
        .flatMap(order -> order.lines().stream())
        .map(LineItem::sku)
        .toList();

The interview trick is to describe the shape coming out. One output per input is map; a stream of outputs per input is flatMap.

Model answer

import java.util.List;

public class Main {
    public static void main(String[] args) {
        int total = List.of(1, 2, 3, 4).parallelStream()
                .reduce(10, Integer::sum);

        System.out.println(total);
    }
}

On a normal Java 21 run this prints:

50

The bug is that 10 is not the identity for addition. In a parallel reduction, the identity can be used once per partition, not once for the whole pipeline. With four one-element partitions, you effectively get (10 + 1) + (10 + 2) + (10 + 3) + (10 + 4), which is 50.

For a real sum, use 0, or better for numbers:

int total = List.of(1, 2, 3, 4).stream()
        .mapToInt(Integer::intValue)
        .sum();