番石榴的选修课有什么意义

我最近读到过关于这个的文章,看到有人在使用这个类,但是在几乎所有的情况下,使用 null也可以很好地工作——如果不是更直观的话。有人能提供一个具体的例子,其中 Optional将实现的东西,null不能或在一个更清洁的方式?我能想到的唯一一件事就是将它与不接受 null键的 Maps一起使用,但即使这样也可以使用 null 值的一侧“映射”来完成。有人能给我一个更有说服力的论据吗?

26218 次浏览

Guava team member here.

Probably the single biggest disadvantage of null is that it's not obvious what it should mean in any given context: it doesn't have an illustrative name. It's not always obvious that null means "no value for this parameter" -- heck, as a return value, sometimes it means "error", or even "success" (!!), or simply "the correct answer is nothing". Optional is frequently the concept you actually mean when you make a variable nullable, but not always. When it isn't, we recommend that you write your own class, similar to Optional but with a different naming scheme, to make clear what you actually mean.

But I would say the biggest advantage of Optional isn't in readability: the advantage is its idiot-proof-ness. It forces you to actively think about the absent case if you want your program to compile at all, since you have to actively unwrap the Optional and address that case. Null makes it disturbingly easy to simply forget things, and though FindBugs helps, I don't think it addresses the issue nearly as well. This is especially relevant when you're returning values that may or may not be "present." You (and others) are far more likely to forget that other.method(a, b) could return a null value than you're likely to forget that a could be null when you're implementing other.method. Returning Optional makes it impossible for callers to forget that case, since they have to unwrap the object themselves.

For these reasons, we recommend that you use Optional as a return type for your methods, but not necessarily in your method arguments.

(This is totally cribbed, by the way, from the discussion here.)

It really looks like the Maybe Monad pattern from Haskell.

You should read the following, Wikipedia Monad (functional programming):

And read From Optional to Monad with Guava on Kerflyn's Blog, which discusses about the Optional of Guava used as a Monad:


Edit: With Java8, there's a built-in Optional that has monadic operators like flatMap. This has been a controversial subject but finally has been implemented.

See http://www.nurkiewicz.com/2013/08/optional-in-java-8-cheat-sheet.html

public Optional<String> tryFindSimilar(String s)  //...


Optional<Optional<String>> bad = opt.map(this::tryFindSimilar);
Optional<String> similar =       opt.flatMap(this::tryFindSimilar);

The flatMap operator is essential to allow monadic operations, and permits to easily chain calls that all return Optional results.

Think about it, if you used the map operator 5 times you would end up with an Optional<Optional<Optional<Optional<Optional<String>>>>>, while using flatMap would give you Optional<String>

Since Java8 I would rather not use Guava's Optional which is less powerful.

The most important advantage of Optional is that it adds more details to the contract between the implementer and caller of a function. For this reason is both useful for parameters and return type.

If you make the convention to always have Optional for possible null objects you add more clarifications to cases like:

  1. Optional<Integer> maxPrime(Optional<Integer> from, Optional<Integer> to)

    The contract here clearly specifies that there is a chance that a result is not returned but also shows that it will work with from and to as absent.

  2. Optional<Integer> maxPrime(Optional<Integer> from, Integer to)

    The contract specifies that the from is optional so an absent value might have a special meaning like start from 2. I can expect that a null value for the to parameter will throw an exception.

So the good part of using Optional is that the contract became both descriptive (similar with @NotNull annotation) but also formal since you must write code .get() to cope with Optional.

One good reason to use it is that it makes your nulls very meaningful. Instead of returning a null that could mean many things (like error, or failure, or empty,etc) you can put a 'name' to your null. Look at this example:

lets define a basic POJO:

class PersonDetails {


String person;
String comments;


public PersonDetails(String person, String comments) {
this.person = person;
this.comments = comments;
}


public String getPerson() {
return person;
}




public String getComments() {
return comments;
}

}

Now lets make use of this simple POJO:

public Optional<PersonDetails> getPersonDetailstWithOptional () {


PersonDetails details = null; /*details of the person are empty but to the caller this is meaningless,
lets make the return value more meaningful*/




if (details == null) {
//return an absent here, caller can check for absent to signify details are not present
return Optional.absent();
} else {
//else return the details wrapped in a guava 'optional'
return Optional.of(details);
}
}

Now lets avoid using null and do our checks with Optional so its meaningful

public void checkUsingOptional () {


Optional<PersonDetails> details = getPersonDetailstWithOptional();


/*below condition checks if persons details are present (notice we dont check if person details are null,
we use something more meaningful. Guava optional forces this with the implementation)*/
if (details.isPresent()) {


PersonDetails details = details.get();


// proceed with further processing
logger.info(details);


} else {
// do nothing
logger.info("object was null");
}


assertFalse(details.isPresent());
}

thus in the end its a way to make nulls meaningful, & less ambiguity.