Functional Programming is nothing new to the JVM. But with Java 8 it was the big change in paradigm for Java developers. Learn how switching from imperative to functional results in more maintainable code. The presentation was given at tech meetup KrkDataLink, The missing link between Functional Programming and Data Science.

1. Functional Programming in Java
Code For Maintainability
1
KrkDataLink
Kraków
2017-02-15
@marcinstepien
www.smart.biz.pl

2. Marcin Stepien
Developer, Consultant
@marcinstepien
www.smart.biz.pl
2005
www.whenvi.com
2014
2

3. What is here
3
Code for maintainability
Functional Programing intro
Object vs Functional
Code examples

4. So the next developer doesn’t hate you
Code For Maintainability
4

5. Maintainable Code
5
Reusable Expressive
Clear > Clever Separate of concerns

6. Closure Lisp
Frege kind of Haskell
Scala
Java 8, Kotlin, Groovy
Functional Programming on
JVM
6

7. Functional
Programming
Java
7

8. ➔ Streams
➔ Lambda expressions
➔ Functions as first class citizens
➔ Functional Interfaces
8
Java 8 FP

9. ➔ No side effects
➔ Final variables
➔ y = f(x) always same y for given x
9
(pure) function

10. Function<Integer,Integer> increase = x -> x + 1;
#parameter -> body
10
Java 8 function

11. 11
Functional interface Function descriptor
Predicate<T> T -> boolean
Consumer<T> T -> void
Function<T,R> T -> R
Supplier<T> () -> T
UnaryOperator<T> T -> T
BinaryOperator<T> (T,T) -> T
BiPredicate<L,R> (L,R) -> boolean
BiConsumer<T,U> (T,U) -> void
BiFunction<T,U,R> (T,U) -> R

12. #instead of
Runnable task =
new Runnable() {
public void run() {
System.out.println(”no arg consumer” )
}
}
12
Replace ceremony with Lambda
Runnable task =
() -> { System.out.println(”no arg consumer” ); };

13. 13
Internal iterators
#instead of loops
for(String name : names) {
System.out.println(name)
}
#use internal iterators
names.forEach(e -> System.out.println(e))
#with method reference
names.forEach(System.out::println)

14. public int getSumOfEvens(List<Integer> nums){
Integer sum = 0;
for(Integer num : nums) {
if(num != null && num % 2 == 0) {
sum += num;
}
}
return sum;
}
14
Fight verbosity with streams
nums.stream()
.filter(Objects::nonNull)
.filter( number -> number % 2 == 0)
.reduce(0, Integer::sum);

15. 15
Passing functions
public int
getSum(List<Integer>nums,Predicate<Integer> filter,
Bifunction<Integer,Integer,Integer> accumulator)
{
return nums.stream()
.filter(filter)
.filter(Objects::nonNull)
.reduce(0, accumulator);
}
Predicate<Integer> isEven = n -> n % 2 == 0;
int sum = getSum(numbers, isEven, Integer::sum);

16. 16
Map reduce, lazy evaluation
public int getTranactions(List<User> users) {
return users.stream() # or parallelStream()
.filter(u -> u.isActivated())
.map( u -> u.getTranTotal())
.reduce(0, Integer::sum);} #terminal op
public Stream<Integer> get(List<User> users) {
#no computation is happening here
return users.stream()
.filter(u -> u.isActivated())
.map( u -> u.getTranTotal());
}

17. 17
Lazy evaluation, infinite series
Stream<Integer> infiniteSeries
= Stream.iterate(1, e -> e + 1)

18. 18
Slimmer patterns: decorator
← https://en.wikipedia.org/wiki/Decorator_pattern#Second_example_.28coffee_making_scenario.29
← Hierarchy of types
can be replaced with Function Composition:
● .compose(Function f)
● .andThan(Function f)

19. 19
Function composition
class Barista {
public static Coffee makeCoffee
(Coffee baseCoffee,
Function<Coffee, Coffee>... addIngredients)
{
return Stream.of(addIngredients)
.reduce(
Function.<Coffee>identity(),
//or Function::andThan
(addIngr1, addIngr2) -> addIngr1.andThan(addIngr2))
Coffee coffee = Barista.makeCoffee(
new Arabica(),
Coffee::withMilk,
Coffee::withSprinkles));

20. Object vs Functional
20
● Imperative
● Mutability
● Lower abstraction
● Reveals impl. details
● Eager by default
● nulls
● Hierarchy of types
● Declarative
● Favors Immutability
● Expressive
● Hides impl. details
● Lazy by default
● Optionals
● Function composition

21. marcin@smart.biz.pl
@marcinstepien
21
Thank you