Skip to main content

Optional in Java 8 cheat sheet

java.util.Optional<T> in Java 8 is a poor cousin of scala.Option[T] and Data.Maybe in Haskell. But this doesn’t mean it’s not useful. If this concept is new to you, imagine Optional as a container that may or may not contain some value. Just like all references in Java can point to some object or be null, Option may enclose some (non-null!) reference or be empty.

Turns out that the analogy between Optional and nullable references is quite sensible. Optional was introduced in Java 8 so obviously it is not used throughout the standard Java library - and never will be for the backward compatibility reasons. But I recommend you at least giving it a try and using it whenever you have nullable references. Optional instead of plain null is statically checked at compile time and much more informative as it clearly indicates that a given variable may be present or not. Of course it requires some discipline - you should never assign null to any variable any more.

Usage of option (maybe) pattern is quite controversial and I am not going to step into this discussion. Instead I present you with few use-cases of null and how they can be retrofitted to Optional<T>. In the following examples given variables and types are used:

public void print(String s) {

String x = //...
Optional<String> opt = //...
x is a String that may be null, opt is never null, but may or may not contain some value (present or empty). There are few ways of creating Optional:

opt = Optional.of(notNull);

opt = Optional.ofNullable(mayBeNull);

opt = Optional.empty();
In the first case Optional must contain not null value and will throw an exception if null is passed. ofNullable() will either return empty or present (set) Optional. empty() always return empty Optional, corresponding to null. It’s a singleton because Optional<T> is immutable.

ifPresent() - do something when Optional is set

Tedious if statement:

if (x != null) {
can be replaced with higher-order function ifPresent():

opt.ifPresent(x -> print(x));
The latter syntax (method reference) can be used when lambda argument (String x) matches function formal parameters.

filter() - reject (filter out) certain Optional values.

Sometimes you want to perform some action not only when a reference is set but also when it meets certain condition:

if (x != null && x.contains("ab")) {
This can be replaced with Optional.filter() that turns present (set) Optional to empty Optional if underlying value does not meet given predicate. If input Optional was empty, it is returned as-is:

   filter(x -> x.contains("ab")).
This is equivalent to more imperative:

if(opt.isPresent() && opt.get().contains("ab")) {

map() - transform value if present

Very often you need to apply some transformation on a value, but only if it’s not null (avoiding NullPointerException):

if (x != null) {
    String t = x.trim();
    if (t.length() > 1) {
This can be done in much more declarative way using map():

    filter(t -> t.length() > 1).
This becomes tricky. applies given function on a value inside Optional - but only if Optional is present. Otherwise nothing happens and empty() is returned. Remember that the transformation is type-safe - look at generics here:

Optional<String>  opt = //...
Optional<Integer> len =;
If Optional<String> is present Optional<Integer> len is present as well, wrapping length of a String. But if opt was empty, map() over it does nothing except changing generic type.

orElse()/orElseGet() - turning empty Optional<T> to default T

At some point you may wish to unwrap Optional and get a hold of real value inside. But you can’t do this if Optional is empty. Here is a pre-Java 8 way of handling such scenario:

int len = (x != null)? x.length() : -1;
With Optional we can say:

int len =;
There is also a version that accepts Supplier<T> if computing default value is slow, expensive or has side-effects:

int len = opt.
    orElseGet(() -> slowDefault());     //orElseGet(this::slowDefault)

flatMap() - we need to go deeper

Imagine you have a function that does not accept null but may produce one:

public String findSimilar(@NotNull String s) //...
Using it is a bit cumbersome:

String similarOrNull = x != null? findSimilar(x) : null;
With Optional it is a bit more straighforward:

Optional<String> similar =;
If the function we map() over returns null, the result of map() is an empty Optional. Otherwise it’s the result of said function wrapped with (present) Optional. So far so good but why do we return null-able value if we have Optional?

public Optional<String> tryFindSimilar(String s)  //...
Our intentions are clear but using map() fails to produce correct type. Instead we must use flatMap():

Optional<Optional<String>> bad =;
Optional<String> similar =       opt.flatMap(this::tryFindSimilar);
Do you see double Optional<Optional<...>>? Definitely not what we want. If you are mapping over a function that returns Optional, use flatMap instead. Here is a simplified implementation of this function:

public <U> Optional<U> flatMap(Function<T, Optional<U>> mapper) {
    if (!isPresent())
        return empty();
    else {
        return mapper.apply(value);

orElseThrow() - lazily throw exceptions on empty Optional

Often we would like to throw an exception if value is not available:

public char firstChar(String s) {
    if (s != null && !s.isEmpty())
        return s.charAt(0);
        throw new IllegalArgumentException();
This whole method can be replaced with the following idiom:

    filter(s -> !s.isEmpty()).
    map(s -> s.charAt(0)).
We don’t want to create an instance of exception in advance because creating an exception has significant cost.

Bigger example

Imagine we have a Person with an Address that has a validFrom date. All of these can be null. We would like to know whether validFrom is set and in the past:

private boolean validAddress(NullPerson person) {
    if (person != null) {
        if (person.getAddress() != null) {
            final Instant validFrom = person.getAddress().getValidFrom();
            return validFrom != null && validFrom.isBefore(now());
        } else
            return false;
    } else
        return false;
Quite ugly and defensive. Alternatively but still ugly:

return person != null &&
       person.getAddress() != null &&
       person.getAddress().getValidFrom() != null &&
Now imagine all of these (person, getAddress(), getValidFrom()) are Optionals of appropriate types, clearly indicating they may not be set:

class Person {

    private final Optional<Address> address;

    public Optional<Address> getAddress() {
        return address;


class Address {
    private final Optional<Instant> validFrom;

    public Optional<Instant> getValidFrom() {
        return validFrom;

Suddenly the computation is much more streamlined:

return person.
        filter(x -> x.before(now())).
Is it more readable? Hard to tell. But at least it’s impossible to produce NullPointerException when Optional is used consistently.

Converting Optional<T> to List<T>

I sometimes like to think about Optional as a collection1 having either 0 or 1 elements. This may make understanding of map() and flatMap() easier. Unfortunately Optional doesn’t have toList() method, but it’s easy to implement one:

public static <T> List<T> toList(Optional<T> option) {
    return option.
Or less idiomatically:

public static <T> List<T> toList(Optional<T> option) {
    if (option.isPresent())
        return Collections.singletonList(option.get());
        return Collections.emptyList();
But why limit ourselves to List<T>? What about Set<T> and other collections? Java 8 already abstracts creating arbitrary collection via Collectors API, introduced for Streams. The API is hideous but comprehensible:

public static <R, A, T> R collect(Optional<T> option, Collector<? super T, A, R> collector) {
    final A container = collector.supplier().get();
    option.ifPresent(v -> collector.accumulator().accept(container, v));
    return collector.finisher().apply(container);
We can now say:

import static*;

List<String> list = collect(opt, toList());
Set<String>  set  = collect(opt, toSet());


Optional<T> is not nearly as powerful as Option[T] in Scala (but at least it doesn’t allow wrapping null). The API is not as straightforward as null-handling and probably much slower. But the benefit of compile-time checking plus readability and documentation value of Optional used consistently greatly outperforms disadvantages. Also it will probably replace nearly identical<T> from Guava

1 - from theoretical point of view both maybe and sequence abstractions are monads, that’s why they share some functionality


  1. Greate art!
    Although you shouldn't use Optional(Address) as class field. See eg. BGoetz note

    1. Brian Goetz did not say so. He only said that "you probably should never use it for something that returns an array of results, or a list of results". And this is only because it is simpler to return an empty array or an empty list.

      Of course, if a property is optional, it should be an Optional and the corresponding getter should return an Optional. What else?

    2. Optional is not Serializable, on purpose:

      It is intended to be used on API level to support the optional-return idiom.

  2. "There is also a version that accepts Supplier if computing default value is slow, expensive or has side-effects:" - I don't understand, how can orElseGet() help with optimisation?

    1. Because the slow method that computes the default will not be called unless necessary. If only orElse() was available, you would have to calculate the default value just to pass it to orElse() even if it will then be discarded.

  3. Hi,

    In the conversion to list example, why not using a method reference, replacing orElse(...) with .orElseGet(Collections::emptyList).

    1. emptyList() returns a singleton so using orElseGet with constant value may actually be faster than Supplier.

    2. Yes, it is faster on the first call. However, orElseGet will not instantiate a Supplier. Using a lambda or a method reference will cause a (very small) overhead on the first call, when a method will be created to represent it. (And method reference is slightly faster, because it does not have to actually create a method, but can use a handle to the actual method instead.) So, if the Optional is empty, it will be slightly slower on the first call, and equivalent on subsequent calls. On the other hand, if the Optional is not empty, the Collections.emptyList() method will be called anyway, so what is supposed to be an optimization will in fact in most cases have worst performance. How much worst depends upon the proportion of calls with an empty Optional. If we suppose that there will be much more calls with a non empty one, the code using a method reference is definitely faster. It also protect your code against future modifications of the emptyList method implementation.

  4. what is checked at compile time if we are using optional?

  5. This comment has been removed by a blog administrator.

  6. What if I want to do something like:

    Optional opt = Optional.empty();
    return opt.isPresent()? opt : (this::of(1));

    i.e. if the opt contains a value, then return it, else stuff it first with a value and return it. Is there any way to manage this?

  7. return opt.isPresent() ? opt : Optional.of(1);


    return Optional.of(opt.orElse(1));

  8. That won't save the value in opt - what if the computation is expensive instead of just Optional.of(1) e.g. Optional.of(findTriillionthPrimeNumber())?

  9. I don't know what you mean by "save the value in opt". If you replace the example by:

    return opt.isPresent() ? opt : Optional.of(getValue());


    return Optional.of(opt.orElse(getValue()));

    in both cases, the getValue() method will be called once if opt is empty. It will not be called otherwise. What more do you want? If you want to cache the value for future accesses because the computation is expensive, it is up to you. But it has nothing to do will Optional.


Post a Comment