I use x != null to avoid NullPointerException. Is there an alternative?
if (x != null) {
// ...
}
How do I avoid checking for nulls in Java?
Asked Answered
This to me sounds like a reasonably common problem that junior to intermediate developers tend to face at some point: they either don't know or don't trust the contracts they are participating in and defensively overcheck for nulls. Additionally, when writing their own code, they tend to rely on returning nulls to indicate something thus requiring the caller to check for nulls.
To put this another way, there are two instances where null checking comes up:
Where null is a valid response in terms of the contract; and
Where it isn't a valid response.
(2) is easy. As of Java 1.7 you can use Objects.requireNonNull(foo). (If you are stuck with a previous version then assertions may be a good alternative.)
"Proper" usage of this method would be like below. The method returns the object passed into it and throws a NullPointerException if the object is null. This means that the returned value is always non-null. The method is primarily intended for validating parameters.
public Foo(Bar bar) {
this.bar = Objects.requireNonNull(bar);
}
It can also be used like an assertion though since it throws an exception if the object is null. In both uses, a message can be added which will be shown in the exception. Below is using it like an assertion and providing a message.
Objects.requireNonNull(someobject, "if someobject is null then something is wrong");
someobject.doCalc();
Generally throwing a specific exception like NullPointerException when a value is null but shouldn't be is favorable to throwing a more general exception like AssertionError. This is the approach the Java library takes; favoring NullPointerException over IllegalArgumentException when an argument is not allowed to be null.
(1) is a little harder. If you have no control over the code you're calling then you're stuck. If null is a valid response, you have to check for it.
If it's code that you do control, however (and this is often the case), then it's a different story. Avoid using nulls as a response. With methods that return collections, it's easy: return empty collections (or arrays) instead of nulls pretty much all the time.
With non-collections it might be harder. Consider this as an example: if you have these interfaces:
public interface Action {
void doSomething();
}
public interface Parser {
Action findAction(String userInput);
}
where Parser takes raw user input and finds something to do, perhaps if you're implementing a command line interface for something. Now you might make the contract that it returns null if there's no appropriate action. That leads the null checking you're talking about.
An alternative solution is to never return null and instead use the Null Object pattern:
public class MyParser implements Parser {
private static Action DO_NOTHING = new Action() {
public void doSomething() { /* do nothing */ }
};
public Action findAction(String userInput) {
// ...
if ( /* we can't find any actions */ ) {
return DO_NOTHING;
}
}
}
Compare:
Parser parser = ParserFactory.getParser();
if (parser == null) {
// now what?
// this would be an example of where null isn't (or shouldn't be) a valid response
}
Action action = parser.findAction(someInput);
if (action == null) {
// do nothing
} else {
action.doSomething();
}
to
ParserFactory.getParser().findAction(someInput).doSomething();
which is a much better design because it leads to more concise code.
That said, perhaps it is entirely appropriate for the findAction() method to throw an Exception with a meaningful error message -- especially in this case where you are relying on user input. It would be much better for the findAction method to throw an Exception than for the calling method to blow up with a simple NullPointerException with no explanation.
try {
ParserFactory.getParser().findAction(someInput).doSomething();
} catch(ActionNotFoundException anfe) {
userConsole.err(anfe.getMessage());
}
Or if you think the try/catch mechanism is too ugly, rather than Do Nothing your default action should provide feedback to the user.
public Action findAction(final String userInput) {
/* Code to return requested Action if found */
return new Action() {
public void doSomething() {
userConsole.err("Action not found: " + userInput);
}
}
}
Nice answer, especially mentioning asserts. I came from C and use them constantly. As to Null Objects, they are NOT a silver bullet. I have often spent hours debugging code which turned out to be a Null Object being used, doing nothing (or returning its defaults) when a NPE would have been indicated the problem much more clearly. So then you'll have to do a "Null Object check" instead of a null check, and haven't won anything, actually lost clarity. Nowadays, if I can't avoid null, I just document it clearly and that's it. –
Publias
Writing
this.bar = Objects.requireNonNull(bar); is a good use of this method, because you're not about to use this.bar. This way, you get the exception when an invalid null object is being passed, rather than when you try to use it. But, it often gets used this way: Objects.requireNonNull(getThing()).doThingStuff(); Here, the call to Objects.requireNonNull() doesn't change the behavior at all, it throws the NPE either way. It is best used when you are saving an object for later use. In my second example, the exception is useful. It helps track down the bug. –
Penultimate Should we suggest using exceptions for control flow though? While it's technically a solution that works, programmers should remember that exceptions should only be used to catch unexpected situations. When validating a user's input, invalid input should ideally not lead to exceptions being thrown. –
Coffeecolored
If you use (or planning to use) a Java IDE like JetBrains IntelliJ IDEA, Eclipse or Netbeans or a tool like findbugs then you can use annotations to solve this problem.
Basically, you've got @Nullable and @NotNull.
You can use in method and parameters, like this:
@NotNull public static String helloWorld() {
return "Hello World";
}
or
@Nullable public static String helloWorld() {
return "Hello World";
}
The second example won't compile (in IntelliJ IDEA).
When you use the first helloWorld() function in another piece of code:
public static void main(String[] args)
{
String result = helloWorld();
if(result != null) {
System.out.println(result);
}
}
Now the IntelliJ IDEA compiler will tell you that the check is useless, since the helloWorld() function won't return null, ever.
Using parameter
void someMethod(@NotNull someParameter) { }
if you write something like:
someMethod(null);
This won't compile.
Last example using @Nullable
@Nullable iWantToDestroyEverything() { return null; }
Doing this
iWantToDestroyEverything().something();
And you can be sure that this won't happen. :)
It's a nice way to let the compiler check something more than it usually does and to enforce your contracts to be stronger. Unfortunately, it's not supported by all the compilers.
In IntelliJ IDEA 10.5 and on, they added support for any other @Nullable @NotNull implementations.
See blog post More flexible and configurable @Nullable/@NotNull annotations.
@NotNull, @Nullable and other nullness annotations are part of JSR 305. You can also use them to detect potential problems with tools like FindBugs. –
Lavery I find it strangely annoying that the
@NotNull & @Nullable interfaces live in the package com.sun.istack.internal. (I guess I associate com.sun with warnings about using a proprietary API.) –
Sophist code portability goes null with jetbrains.I would think twice(square) before tying down to ide level.Like Jacek S said they are part of JSR any way which I thought was JSR303 by the way. –
Indulgence
Why doesn't the 2nd one compile though? –
Virus
I really don't think that using a custom compiler is a viable solution to this problem. –
Valerivaleria
The good thing about annotations, which
@NotNull and @Nullable are is that they nicely degrade when the source code is built by a system that doesn't understand them. So, in effect, the argument that the code is not portable may be invalid - if you use a system that supports and understands these annotations, you get the added benefit of stricter error checking, otherwise you get less of it but your code should still build fine, and the quality of your running program is THE SAME, because these annotations were not enforced at runtime anyway. Besides, all compilers are custom ;-) –
Imperforate in your example you call
helloWorld twice. This is terrible. you have no idea what could be the consequences ! –
Advisable @Mechanicalsnail becuase it only ever returns "hello world", and can never return null. –
Thenceforth
@JacekS Just for information - for Android development, As of version 19.1 of the Android support library,
@NotNull and @Nullable annotation support has been added. So basically all Android developer can take advantage of it. It won't be IDE specific anymore. More information regarding support annotations here - tools.android.com/tech-docs/support-annotations –
Homeroom Arguments against
@Nullable/@NotNull for "compatibility" are missing the point. They are only there to help tools (from a static analysis) ensure that these basic contracts are met - they do not replace proper null handling, when such is appropriate per the contract (or possible per the input source). While it would be awesome if Java and C# (and the run-times) actually supported not-null reference types in the Type System, using better static analysis tooling (and explicit annotated contracts) goes a long way to reduce the 100 trillion dollar mistake that is Algo, is Java, is C# .. –
Anachronism Netbeans also has support for
@NotNull and @Nullable. –
Knapweed It's worth noting that TYPE_USE annotations (since Java 8) let you treat null annotations as an integral part of the type system, including in particular generics. This means that conceptually all contract violations can be detected by the compiler. Eclipse supports this since Java 8 was released, see help.eclipse.org/topic/org.eclipse.jdt.doc.user/tasks/… –
Capitalize
This should now be the primary answer, I believe the accepted answer was accepted before these annotations were available. I wonder if there is a way to get it promoted to the top at least... –
Millian
By the way, the implementations of the checkers are not all consistent! Eclipse's checker is a good deal better than IntelliJ's (although I've been using IntelliJ's, it's definitely better than nothing). Eclipses has much better handling of un-annotated variables. stackoverflow.com/questions/48349393/… –
Millian
In the second example, IntelliJ tells you that you're applying the
@Nullable annotation to a method that will never return null. –
Penultimate @JavaKaBaby Code portability does NOT suffer with JetBrains. All JetBrains does is make NullPointerExceptions a compile-time error instead of a runtime error. Building with another compiler won't change the behavior of the code at all, it just delays the reporting of the bug until runtime. The code ports just fine. This is an understandable gut-reaction to any suggestion to use some compiler-specific solution, but in this case, when you think it through, you'll see that this is still a viable option. And now that other compilers support this, it's becoming even more viable. –
Penultimate
If null-values are not allowed
If your method is called externally, start with something like this:
public void method(Object object) {
if (object == null) {
throw new IllegalArgumentException("...");
}
Then, in the rest of that method, you'll know that object is not null.
If it is an internal method (not part of an API), just document that it cannot be null, and that's it.
Example:
public String getFirst3Chars(String text) {
return text.subString(0, 3);
}
However, if your method just passes the value on, and the next method passes it on etc. it could get problematic. In that case you may want to check the argument as above.
If null is allowed
This really depends. If find that I often do something like this:
if (object == null) {
// something
} else {
// something else
}
So I branch, and do two completely different things. There is no ugly code snippet, because I really need to do two different things depending on the data. For example, should I work on the input, or should I calculate a good default value?
It's actually rare for me to use the idiom "if (object != null && ...".
It may be easier to give you examples, if you show examples of where you typically use the idiom.
What's the point in throwing IllegalArgumentException? I think NullPointerException would be clearer, and that would also be thrown if you don't do the null-check yourself. I'd either use assert or nothing at all. –
Frontality
It's unlikely that every other value than null is acceptable. You could have IllegalArgumentException, OutOfRageException etc etc. Sometimes this makes sense. Other times you end up creating a lot of exception classes that doesn't add any value, then you just use IllegalArgumentException. It doesn't makes sense to have one exception for null-input, and another one for everything else. –
Hemianopsia
Yes, I agree to fail-fast-principle, but in the example given above, the value is not passed on, but is the object on which a method shall be called. So it fails equallys fast, and adding a null check just to throw an exception that would be thrown anyway at the same time and place doesn't seem to make debugging any easier. –
Frontality
I agree with @Frontality in that IllegalArgumentException implies (as I use it anyway) that a precondition for the argument does not hold as it pertains to a constraint on the parameter. A null parameter is usually just "a general error in programming" and not an application/API specific constraint. To maintain a defensive fail-fast strategy, there's nothing stopping anyone for checking for a null and throwing a NullPointerException... which exactly conveys what happened, namely, "you didn't program it right" :) –
Dildo
"A null parameter is usually just "a general error in programming" and not an application/API specific constraint." Naive. You can use this as a security hole to DoS or attempt to gain stacktrace info (if such information is not supressed). osvdb.org/show/osvdb/67379 If your API is public and you don't explicitly define the boundaries you accept--both through API docs and whatever tools you can use to ensure validity of the contract-- the caller's free to do whatever they want. You are the one who didn't program right then, not the caller. –
Gluten
A security hole? The JDK is full of code like this. If you don't want the user to see stacktraces, then just disable them. No one implied that the behaviour is not documented. MySQL is written in C where dereferencing null pointers is undefined behaviour, nothing like throwing an exception. –
Crosscut
throw new IllegalArgumentException("object==null") –
Kala @Hemianopsia I think it does make sense to have a different exception for null values passed in that from improper values passed in. The two are usually results of different types of programming errors. Null values usually arise from some required computation not happening, while other problems usually arise from required computations being incorrect. –
Foreshow
@Frontality the whole point of using anything other than RuntimeException is the descriptiveness of the Exception. Why IllegalArgumentException over NullPointerException? Because when you are debugging an exception caused by this, the IllegalArgumentException is abundantly more obvious, especially when you put a message on it. This especially applies to enterprise applications when NullPointerExceptions creep up (yes, these happen in enterprise applications). In this same example, the programmer may not have control over whether or not assertions get processed by the JVM. –
Wells
1/2 I have to agree that a
IllegalArgumentException is better than a plain-vanilla NPE for these type of situations. I see a NPE as an indication that somewhere in the code, something unsafely de-referenced an expression that happened to be null (assuming all known and declared preconditions and invariants were met). An illegal argument exception OTH tells me a well-known precondition or invariant was not met. –
Presnell 2/2 - That is, a NPE tells me there is a yet-undocumented null bug within the method (or within the methods it calls) or that there is a null precondition that we have not coded for. An illegal argument exception tells there is a bug outside of this method lying somewhere along the call chain that leads to the current method. Sticking to this pattern (within reason obviously), it does lead to faster root-cause analysis of errors (specially the nasty, intertwined ones.) –
Presnell
Throwing
NullPointerException is the Java convention. See e.g. Objects.requireNonNull which is "primarily for doing parameter validation in methods and constructors" and throws NullPointerException. There is also simply the doc for NullPointerException which states that "applications should throw instances of this class to indicate other illegal uses of the null object". –
Carreon Agree with @Presnell on the differential uses of NPE and IAE. However, this suggests - to me at least - that Java is really lacking a
NullArgumentException as a subclass of IllegalArgumentException. –
Brozak Wow, I almost hate to add another answer when we have 57 different ways to recommend the NullObject pattern, but I think that some people interested in this question may like to know that there is a proposal on the table for Java 7 to add "null-safe handling"—a streamlined syntax for if-not-equal-null logic.
The example given by Alex Miller looks like this:
public String getPostcode(Person person) {
return person?.getAddress()?.getPostcode();
}
The ?. means only de-reference the left identifier if it is not null, otherwise evaluate the remainder of the expression as null. Some people, like Java Posse member Dick Wall and the voters at Devoxx really love this proposal, but there is opposition too, on the grounds that it will actually encourage more use of null as a sentinel value.
Update: An official proposal for a null-safe operator in Java 7 has been submitted under Project Coin. The syntax is a little different than the example above, but it's the same notion.
Update: The null-safe operator proposal didn't make it into Project Coin. So, you won't be seeing this syntax in Java 7.
I think this is wrong. There should be a way to specify that a given variable is ALWAYS non-null. –
Kala
Update: the proposal will not make Java7. See blogs.sun.com/darcy/entry/project_coin_final_five . –
Berthaberthe
But on the same page, see the link to types.cs.washington.edu/jsr308 for info about using the Checker framework to add the @Nullable and @NonNull type annotations in Java 7. –
Nestling
Interesting idea but the choice of syntax is absurd; I don't want a codebase full of question marks pinned into every joint. –
Diderot
This operator exists in Groovy, so those who want to use it still have that as an option. –
Cidevant
Optional was added to Java8 –
Rabble
This is the most ingenious idea I've seen. It should be added to every sensible language of the C syntax. I'd rather "pin question marks" everywhere than scroll through screenful of lines or dodge "guard clauses" all day. –
Reeding
Tony Hoare apologised for inventing
null. But an even bigger mistake is to pick a strategy, to stick to it for over 20 years until billions lines of code are written, to then finally change your mind. This .? was like a last chance to keep java "clean". It's a very simple syntactical solution. But instead they prefer to throw in a handful of different @Nullable and Option classes, empty collections, empty objects. The goal was to make nullability easier. But instead they added many layers of complexity. In the end, this will split the java community and will drive it to distruction. –
Helvetian Kotlin has the null safe call operators, too. kotlinlang.org/docs/reference/null-safety.html#safe-calls –
Meatball
@Helvetian I totally agree (except of your criticism of empty objects, because these things exit). Also, no need to apologize for "null" IMO -- to me, it's a totally natural feature of references, and I'm okay with it. I'm absolutely not OK with Oracle refusing to deal with it -- the ".?" syntax is so neat and concise and clear, I really don't get why they persist in providing baroque, cumbersome, ineffective pseudo-solutions. –
Publias
@FranzD. I've wrote a more elaborate post about this here: stackoverflow.com/questions/26529091/… –
Helvetian
If undefined values are not permitted:
You might configure your IDE to warn you about potential null dereferencing. E.g. in Eclipse, see Preferences > Java > Compiler > Errors/Warnings/Null analysis.
If undefined values are permitted:
If you want to define a new API where undefined values make sense, use the Option Pattern (may be familiar from functional languages). It has the following advantages:
- It is stated explicitly in the API whether an input or output exists or not.
- The compiler forces you to handle the "undefined" case.
- Option is a monad, so there is no need for verbose null checking, just use map/foreach/getOrElse or a similar combinator to safely use the value (example).
Java 8 has a built-in Optional class (recommended); for earlier versions, there are library alternatives, for example Guava's Optional or FunctionalJava's Option. But like many functional-style patterns, using Option in Java (even 8) results in quite some boilerplate, which you can reduce using a less verbose JVM language, e.g. Scala or Xtend.
If you have to deal with an API which might return nulls, you can't do much in Java. Xtend and Groovy have the Elvis operator ?: and the null-safe dereference operator ?., but note that this returns null in case of a null reference, so it just "defers" the proper handling of null.
Indeed, the Option pattern is awesome. Some Java equivalents exist. Guava contains a limited version of this called Optional which leaves out most of the functional stuff. In Haskell, this pattern is called Maybe. –
Rajasthan
@Luca Molteni: You're absolutely right, I already commented on the post to request expanding it. :) –
Phyllotaxis
An Optional class will be available in Java 8 –
Tatiana
...and it doesn't (yet) have map nor flatMap: download.java.net/jdk8/docs/api/java/util/Optional.html –
Phyllotaxis
The Optional pattern doesn't solve anything; instead of one potentially null object, now you have two. –
Antipode
Also, look at Functional Guava Extensions (fugue). This library also has its own Option. –
Sherbrooke
Only for this situation -
Not checking if a variable is null before invoking an equals method (a string compare example below):
if ( foo.equals("bar") ) {
// ...
}
will result in a NullPointerException if foo doesn't exist.
You can avoid that if you compare your Strings like this:
if ( "bar".equals(foo) ) {
// ...
}
I agree - only in that situation. I can't stand programmers that have taken this to the next unnecessary level and write if (null != myVar)... just looks ugly to me and serves no purpose! –
Dimerous
This is a particular example, probably the most used, of a general good practice: if you know it, always do
<object that you know that is not null>.equals(<object that might be null>);. It works for other methods other than equals if you know the contract and those methods can handle null parameters. –
Tipper This is the first example I have seen of Yoda Conditions that actually makes sense –
Velazquez
NullPointerExceptions are thrown for a reason. They are thrown because an object is null where it shouldn't be. It is the programmers job to FIX this, not HIDE the problem. –
Defoe
This only hides the problem. What if you are using
foo later on? If a variable shouldn't be null, but is...your app needs to get a NPE so you as the developer can actually fix the underlying reason. –
Birecree Java 8 introduced the java.util.Optional class that addresses some of the problem. One can at least say that it improves the readability of the code, and in the case of public APIs make the contract clearer to the client developer.
They work as follows:
An optional object for a given type (Fruit) is created as the return type of a method. It can be empty or contain a Fruit object:
public static Optional<Fruit> find(String name, List<Fruit> fruits) {
for (Fruit fruit : fruits) {
if (fruit.getName().equals(name)) {
return Optional.of(fruit);
}
}
return Optional.empty();
}
Now look at this code where we search a list of Fruit (fruits) for a given Fruit instance:
Optional<Fruit> found = find("lemon", fruits);
if (found.isPresent()) {
Fruit fruit = found.get(); // Prefer found.orElseThrow() in Java 9+
String name = fruit.getName();
}
You can use the map method to perform a computation on — or extract a value from — an optional object. The orElse method lets you provide a fallback for missing values, converting an Optional to a non-nullable value.
String nameOrFallback = find("lemon", fruits)
.map(f -> f.getName())
.orElse("empty-name");
Of course, handling the null/empty value is still necessary, but at least the developer is conscious that the value might be empty and the risk of forgetting to check is limited.
In an API built from scratch using Optional whenever a return value might be empty, and returning a plain object only when it cannot be null (convention), the client code might abandon null checks on simple object return values.
Of course Optional could also be used as a method argument, perhaps a better way to indicate optional arguments than 5 or 10 overloading methods in some cases. Be aware that using it in this manner is recommended against by Java Language Architect Brian Goetz: "You should almost never use Optional as […] a method parameter."
Optional offers other convenient methods, such as orElseGet that calculates a default value only if the Optional is empty, and ifPresent that peforms an operation only if the Optional is non-empty, both of which work with lambda expressions.
I invite you to read this article (the main source for writing this answer) in which the problematic behavior of NullPointerException (and in general null pointer) and the (partial) solution brought by Optional are well explained: Java Optional Objects.
Google's guava has an optional implimention for Java 6+. –
Habilitate
It's very important to emphasise that using Optional only with ifPresent() does not add much value above normal null checking. It's core value is that it is a monad that can be used in function chains of map/flapMap, which achieves results similar to the Elvis operator in Groovy mentioned elsewhere. Even without this usage, though, I find the orElse/orElseThrow syntax also very useful. –
Overt
This blog has a good entry on Optional winterbe.com/posts/2015/03/15/avoid-null-checks-in-java –
Pisolite
I really never understood why people are so happy with this boilerplate code dzone.com/articles/java-8-elvis-operator –
Agnes
Why people have tendency to do this
if(optional.isPresent()){ optional.get(); } instead of optional.ifPresent(o -> { ...}) –
Geotropism So, other than the API contractual hints, it's really just about catering to functional programmers who enjoy chaining methods endlessly. –
Moneyer
I know this post is old, but above example is a wrong use of Java 8's features imho. Do not combine imperative code with optionals like that. When having a fruit list + a name to search for, use the Streaming API:
fruits.stream().filter(f -> name.equals(f.getName()).findFirst().map(Fruit::getName).orElse("empty-name");. –
Tibbs Depending on what kind of objects you are checking you may be able to use some of the classes in the apache commons such as: apache commons lang and apache commons collections
Example:
String foo;
...
if( StringUtils.isBlank( foo ) ) {
///do something
}
or (depending on what you need to check):
String foo;
...
if( StringUtils.isEmpty( foo ) ) {
///do something
}
The StringUtils class is only one of many; there are quite a few good classes in the commons that do null safe manipulation.
Here follows an example of how you can use null vallidation in JAVA when you include apache library(commons-lang-2.4.jar)
public DOCUMENT read(String xml, ValidationEventHandler validationEventHandler) {
Validate.notNull(validationEventHandler,"ValidationHandler not Injected");
return read(new StringReader(xml), true, validationEventHandler);
}
And if you are using Spring, Spring also has the same functionality in its package, see library(spring-2.4.6.jar)
Example on how to use this static classf from spring(org.springframework.util.Assert)
Assert.notNull(validationEventHandler,"ValidationHandler not Injected");
Also you can use the more generic version from Apache Commons, quite useful at the start of methods to check params I find. Validate.notNull( object, "object must not be null"); commons.apache.org/lang/apidocs/org/apache/commons/lang/… –
Walkover
@Walkover does Validate use Assert statements into?. i ask that because Assert may be activate / deactivate on JVM and it's suggest do not use in production. –
Firehouse
Don't think so - I believe it just throws a RuntimeException if validation fails –
Walkover
- If you consider an object should not be null (or it is a bug) use an assert.
- If your method doesn't accept null params say it in the javadoc and use an assert.
You have to check for object != null only if you want to handle the case where the object may be null...
There is a proposal to add new annotations in Java7 to help with null / notnull params: http://tech.puredanger.com/java7/#jsr308
No, do not use assertions in production code. –
Jesse
I'm a fan of "fail fast" code. Ask yourself - are you doing something useful in the case where the parameter is null? If you don't have a clear answer for what your code should do in that case... i.e. - it should never be null in the first place, then ignore it and allow a NullPointerException to be thrown. The calling code will make just as much sense of an NPE as it would an IllegalArgumentException, but it'll be easier for the developer to debug and understand what went wrong if an NPE is thrown rather than your code attempting to execute some other unexpected contingency logic - which ultimately results in the application failing anyway.
better to use assertions, i.e Contract.notNull(abc, "abc must be non-null, did it fail to load during xyz?"); - this is a more compact way than doing an if (abc!=null) { throw new RuntimeException...} –
Baseboard
Sometimes, you have methods that operate on its parameters that define a symmetric operation:
a.f(b); <-> b.f(a);
If you know b can never be null, you can just swap it. It is most useful for equals:
Instead of foo.equals("bar"); better do "bar".equals(foo);.
But then you have to assume
equals (could be any method) will handle null correctly. Really all this is doing is passing the responsibility to someone else (or another method). –
Kalk @Kalk Basically yes, but
equals (or whatever method) has to check for null anyway. Or state explicitly that it does not. –
Egyptology Rather than Null Object Pattern -- which has its uses -- you might consider situations where the null object is a bug.
When the exception is thrown, examine the stack trace and work through the bug.
Problem is that usually you loose context as the NullPointerException does not indicate WHICH variable was null, and you may have several "."-operations on the line. Using "if (foo == null) throw new RuntimeException("foo == null")" allows you to state explicitly WHAT was wrong, giving your stack trace much more value to those who have to fix it. –
Kala
With Andersen - I would love Java's exception system to be able to include the name of a variable that's being worked upon, so that NullPointerExceptions would not only indicate the line the exception occurred on, but also the variable name. This should work just fine in unobfuscated software. –
Olenolin
I had a professor that preached against method call chaining. His theory was that you should be wary of call chains that were longer than 2 methods. I don't know if that's a hard rule, but it definitely removes most of the problems with NPE stack traces. –
Hardenberg
@Hardenberg You could just put each call on the chain on its own line. That way you can still chain until the cows come home, but you know exactly where things went awry. –
Zuckerman
In java17 the error-message of a NullPointerException finally includes some important details in the message: dev.to/harsvnc/the-new-nullpointer-exception-in-java-17-4bcb –
Vinegarish
The Google collections framework offers a good and elegant way to achieve the null check.
There is a method in a library class like this:
static <T> T checkNotNull(T e) {
if (e == null) {
throw new NullPointerException();
}
return e;
}
And the usage is (with import static):
...
void foo(int a, Person p) {
if (checkNotNull(p).getAge() > a) {
...
}
else {
...
}
}
...
Or in your example:
checkNotNull(someobject).doCalc();
mmm, what is the difference? p.getAge() would throw the same NPE with less overhead and a clearer stack trace. What am I missing? –
Kirkland
It is better to throw an IllegalArgumentException("e == null") in your example as it clearly indicates that it is a programmer-intended exception (along with enough information to actually allow the maintainer to identify the problem). NullPointerExceptions should be reserved for the JVM, as it then clearly indicates that this was unintentional (and usually happens somewhere hard to identify) –
Kala
This is now part of Google Guava. –
Microcircuit
Smells like over-engineering to me. Just let the JVM throw an NPE and don't clutter your code with this junk. –
Dimerous
I like it and open most methods and constructors with explicit checks on the arguments; if there is an error, methods always fail on the first few lines and I know the offending reference without finding something like
getThing().getItsThing().getOtherThing().wowEncapsulationIsBroken().setLol("hi"); –
Ayeaye Java 1.7 added the
java.util.Objects.requireNonNull method (amongst others) that does exactly this. It's such a common idiom that I'm amazed it took this long! –
Opportunism just for interest... what is the difference between checkNotNull(bla).foo() and bla.foo() if both may throw a NPE in case of null? zero benefits, right? –
Capercaillie
Null is not a 'problem'. It is an integral part of a complete modeling tool set. Software aims to model the complexity of the world and null bears its burden. Null indicates 'No data' or 'Unknown' in Java and the like. So it is appropriate to use nulls for these purposes. I don't prefer the 'Null object' pattern; I think it rise the 'who will guard
the guardians' problem.
If you ask me what is the name of my girlfriend I'll tell you that I have no girlfriend. In the Java language I'll return null.
An alternative would be to throw meaningful exception to indicate some problem that can't be (or don't want to be) solved right there and delegate it somewhere higher in the stack to retry or report data access error to the user.
For an 'unknown question' give 'unknown answer'. (Be null-safe where this is correct from business point of view) Checking arguments for null once inside a method before usage relieves multiple callers from checking them before a call.
public Photo getPhotoOfThePerson(Person person) { if (person == null) return null; // Grabbing some resources or intensive calculation // using person object anyhow. }Previous leads to normal logic flow to get no photo of a non-existent girlfriend from my photo library.
getPhotoOfThePerson(me.getGirlfriend())And it fits with new coming Java API (looking forward)
getPhotoByName(me.getGirlfriend()?.getName())While it is rather 'normal business flow' not to find photo stored into the DB for some person, I used to use pairs like below for some other cases
public static MyEnum parseMyEnum(String value); // throws IllegalArgumentException public static MyEnum parseMyEnumOrNull(String value);And don't loathe to type
<alt> + <shift> + <j>(generate javadoc in Eclipse) and write three additional words for you public API. This will be more than enough for all but those who don't read documentation./** * @return photo or null */or
/** * @return photo, never null */This is rather theoretical case and in most cases you should prefer java null safe API (in case it will be released in another 10 years), but
NullPointerExceptionis subclass of anException. Thus it is a form ofThrowablethat indicates conditions that a reasonable application might want to catch (javadoc)! To use the first most advantage of exceptions and separate error-handling code from 'regular' code (according to creators of Java) it is appropriate, as for me, to catchNullPointerException.public Photo getGirlfriendPhoto() { try { return appContext.getPhotoDataSource().getPhotoByName(me.getGirlfriend().getName()); } catch (NullPointerException e) { return null; } }Questions could arise:
Q. What if
getPhotoDataSource()returns null?
A. It is up to business logic. If I fail to find a photo album I'll show you no photos. What if appContext is not initialized? This method's business logic puts up with this. If the same logic should be more strict then throwing an exception it is part of the business logic and explicit check for null should be used (case 3). The new Java Null-safe API fits better here to specify selectively what implies and what does not imply to be initialized to be fail-fast in case of programmer errors.Q. Redundant code could be executed and unnecessary resources could be grabbed.
A. It could take place ifgetPhotoByName()would try to open a database connection, createPreparedStatementand use the person name as an SQL parameter at last. The approach for an unknown question gives an unknown answer (case 1) works here. Before grabbing resources the method should check parameters and return 'unknown' result if needed.Q. This approach has a performance penalty due to the try closure opening.
A. Software should be easy to understand and modify firstly. Only after this, one could think about performance, and only if needed! and where needed! (source), and many others).PS. This approach will be as reasonable to use as the separate error-handling code from "regular" code principle is reasonable to use in some place. Consider the next example:
public SomeValue calculateSomeValueUsingSophisticatedLogic(Predicate predicate) { try { Result1 result1 = performSomeCalculation(predicate); Result2 result2 = performSomeOtherCalculation(result1.getSomeProperty()); Result3 result3 = performThirdCalculation(result2.getSomeProperty()); Result4 result4 = performLastCalculation(result3.getSomeProperty()); return result4.getSomeProperty(); } catch (NullPointerException e) { return null; } } public SomeValue calculateSomeValueUsingSophisticatedLogic(Predicate predicate) { SomeValue result = null; if (predicate != null) { Result1 result1 = performSomeCalculation(predicate); if (result1 != null && result1.getSomeProperty() != null) { Result2 result2 = performSomeOtherCalculation(result1.getSomeProperty()); if (result2 != null && result2.getSomeProperty() != null) { Result3 result3 = performThirdCalculation(result2.getSomeProperty()); if (result3 != null && result3.getSomeProperty() != null) { Result4 result4 = performLastCalculation(result3.getSomeProperty()); if (result4 != null) { result = result4.getSomeProperty(); } } } } } return result; }PPS. For those fast to downvote (and not so fast to read documentation) I would like to say that I've never caught a null-pointer exception (NPE) in my life. But this possibility was intentionally designed by the Java creators because NPE is a subclass of
Exception. We have a precedent in Java history whenThreadDeathis anErrornot because it is actually an application error, but solely because it was not intended to be caught! How much NPE fits to be anErrorthanThreadDeath! But it is not.Check for 'No data' only if business logic implies it.
public void updatePersonPhoneNumber(Long personId, String phoneNumber) { if (personId == null) return; DataSource dataSource = appContext.getStuffDataSource(); Person person = dataSource.getPersonById(personId); if (person != null) { person.setPhoneNumber(phoneNumber); dataSource.updatePerson(person); } else { Person = new Person(personId); person.setPhoneNumber(phoneNumber); dataSource.insertPerson(person); } }and
public void updatePersonPhoneNumber(Long personId, String phoneNumber) { if (personId == null) return; DataSource dataSource = appContext.getStuffDataSource(); Person person = dataSource.getPersonById(personId); if (person == null) throw new SomeReasonableUserException("What are you thinking about ???"); person.setPhoneNumber(phoneNumber); dataSource.updatePerson(person); }If appContext or dataSource is not initialized unhandled runtime NullPointerException will kill current thread and will be processed by Thread.defaultUncaughtExceptionHandler (for you to define and use your favorite logger or other notification mechanizm). If not set, ThreadGroup#uncaughtException will print stacktrace to system err. One should monitor application error log and open Jira issue for each unhandled exception which in fact is application error. Programmer should fix bug somewhere in initialization stuff.
Just because it was intended to mean "no data" doesn't mean it should be used. The language implementation of
null sucks. –
Virus Catching
NullPointerException and returning null is horrible to debug. You end up with NPE later on anyway, and it's really hard to figure out what was originally null. –
Bestial The problem with your example is that getPhotoOfThePerson really has no business deciding whether or not to return a photo or how to deal with a null person -- a straightforward reading of the method name indicates that it returns a photo of a person. It shouldn't return null because it doesn't (and shouldn't) know how the caller will handle a null photo. The caller on the other hand should know what to do when the person is null. Java doesn't have null propogation, which would eliminate the problem, as then it would know enough about how nulls are handled to go ahead and use it. –
Sigil
I'd downvote if I had the reputation. Not only is null not necessary, it's a hole in the type system. Assigning a Tree to a List is a type error because trees are not values of type List; by that same logic, assigning null should be a type error because null is not a value of type Object, or any useful type for that matter. Even the man that invented null considers it his "billion-dollar mistake". The notion of "a value that might be a value of type T OR nothing" is its own type, and should be represented as such (e.g. Maybe<T> or Optional<T>). –
Beckford
As of "Assigning a Tree to a List" you were not talking about Java, did you? Because Java is 'compile time strict type' language (it's memory model differ from C++, C# though). Java compiler will not give you an opportunity to assign Tree to a List. –
Agnes
As of "Maybe<T> or Optional<T>" you still need to write code like
if (maybeNull.hasValue()) {...} so what is the difference with if (maybeNull != null)) {...}? –
Agnes As of "Java doesn't have null propagation, which would eliminate the problem", usually propagation is about 'exceptional cases'. But data availability or non-availability is a common business flow in most cases. If it is something exceptional in data non-availability (which is dictated by business) you should explicitly throw reasonable exception with a reasonable message. Null propagation (with line number inside?..) would be really horrible as for me. –
Agnes
As of "catching NullPointerException and returning null is horrible to debug. You end up with NPE later on anyway, and it's really hard to figure out what was originally null". I'm totally agree! In those cases you should write a dozen of 'if' statements or throw NPE if business logic imply data in-place, or use null-safe operator from new Java. But there are cases when I don't care about what exact step give me null. For example calculating some values for the user just before showing on the screen when you do expect data could be missing. –
Agnes
@MykhayloAdamovych: The benefit of
Maybe<T> or Optional<T> isn't in the case where your T might be null, but in the case where it should never be null. If you have a type that explicitly means "this value might be null -- use with caution", and you use and return such a type consistently, then whenever you see a plain old T in your code, you can assume it is never null. (Course, this would be a lot more useful if enforceable by the compiler.) –
Vasculum Agree. But you'll not reduce 'checks for emptiness' in your code (the topic we are discussing), you'll be more safe though but it comes at a price. –
Agnes
"Even the man that invented null considers it his "billion-dollar mistake" - null was there before anything was invented. One can invent set of rules to deal with it in specific way but this will be 'just another functionally complete system'. I would like to know how much would it cost to deal with subtle bugs in such non fail fast system. May be we are close to compare such systems because of new trends. –
Agnes
"The notion of "a value that might be a value of type T OR nothing" is its own type, and should be represented as such (e.g. Maybe<T> or Optional<T>)" - you may think that all references in Java are Optional<T> and you may switch to DefinetlyInitialized<T> by for using underused final keyword or introduce Mandatory<T>. So what's the difference?.. Anyway I like current Java defaults more because things around us are rather optional. It is optional for 'the kettle to be filled with water before cooking', and that 'there is some fuel in the car before you leave' etc. –
Agnes
Currently reading some Oracle Database Reference about 'Null Values in Arithmetic Expressions'. It states: 'If any column value in an arithmetic expression is null, the result is null. For example, if you attempt to perform division by zero, you get an error. However, if you divide a number by null, the result is a null or unknown.' But don't mind, this is totally not logical, there should be some silly guy somewhere at Oracle and he must be fired till these days... –
Agnes
@MykhayloAdamovych: If meaningful values for array items will become known in arbitrary sequence, and it will be necessary to read some array items before meaningful values have been assigned to all of them, it will generally not be possible to statically rule out the possibility that items will be read before a meaningful value has been determined. If Hoare doesn't like nulls, I'm curious what kind of default value or behavior he'd prefer? Trapping when a null is read from the array might sometimes be better than waiting until it is dereferenced, but would make it impossible to copy... –
Lovage
...a partially-populated array by iteratively copying the items therein. The only other choices I can see is to say that attempting to dereference a pointer read from an array slot that hasn't yet been written will access some default object (which would for most purposes seem rather silly) or will have C-style "Undefined Behavior". I'd say trapping null pointers are better than any alternative I can see. –
Lovage
@MykhayloAdamovych see how Swift does it, which is how many others did before. If Java simply didn't have null, you'd have no way to store a null into a field/variable; to handle the case where it's optional, you'd have to have an Optional<T>. And you ask - what would that gain? Wouldn't you need as many checks? No, you wouldn't, because the compiler would ensure that every non-Optional was never null, and more than half our checks are with objects that never should be null and never are, or if they are then that's a bug that would either be caught by the compiler or exposed by the interface. –
Uniseptate
@Uniseptate i don't follow, After the compiler ensures that the Optional (or what ever) is not null then what? Are you saying that the Person application is designed to never encounter a situation where someone does not have a girlfriend? How does the application handle those scenarios, are there multiple types of Person, maybe we create 2 types of Person, LonelyPerson and ShackledPerson? –
Suzannsuzanna
@entonio, Java has Optional<T>, that is non-final reference. And Java's compiler ensures that every final is initialized. And you would have the same amount of checks if you are senior enough not to write boilerplate code because 'emptiness' is a business notion. And Java's Optional<T> is free from
get() and put() boilerplate. –
Agnes Java 7 has a new java.util.Objects utility class on which there is a requireNonNull() method. All this does is throw a NullPointerException if its argument is null, but it cleans up the code a bit. Example:
Objects.requireNonNull(someObject);
someObject.doCalc();
The method is most useful for checking just before an assignment in a constructor, where each use of it can save three lines of code:
Parent(Child child) {
if (child == null) {
throw new NullPointerException("child");
}
this.child = child;
}
becomes
Parent(Child child) {
this.child = Objects.requireNonNull(child, "child");
}
Actually, your example constitutes code bloat: the first line is superfluous because the NPE would be thrown in the second line. ;-) –
Mclean
True. A better example would be if the second line were
doCalc(someObject). –
Horned Depends. If you are the author of doCalc(), I'd suggest putting the check into that method's body (if possible). And then you most likely will call someObject.someMethod() where again there is no need to check for null. :-) –
Mclean
Well, if you are not the author of
doCalc(), and it doesn't immediately throw NPE when given null, you'd need to check for null and throw NPE yourself. That's what Objects.requireNonNull() is for. –
Horned It's not just code bloat. Better to check up front than halfway through a method that causes side effects or uses time/space. –
Steadman
I probably wouldn't have thrown an NPE, when I'm argument checking for nulls I usually throw an IllegalArgumentException, and it's not a bad thing to do at the very top, if you assume this.child = child will "catch" it, you aren't being explicit in your code, a refactor could make it really tough to catch where the null originated. Code bloat is much better than magic, any day. –
Millian
The
Objects.requireNonNull(object, string) version (where you can pass an error message as 2nd argument) is more useful than the single-arg version (the latter probably just added for the sake of completeness), especially when you're using it to check multiple argument values. Say you had to check 3 arguments for null, this would reduce your code from 3 if/then-throw blocks (separated by empty lines), to 3 single lines not needing empty lines. –
Overt Ultimately, the only way to completely solve this problem is by using a different programming language:
- In Objective-C, you can do the equivalent of invoking a method on
nil, and absolutely nothing will happen. This makes most null checks unnecessary, but it can make errors much harder to diagnose. - In Nice, a Java-derived language, there are two versions of all types: a potentially-null version and a not-null version. You can only invoke methods on not-null types. Potentially-null types can be converted to not-null types through explicit checking for null. This makes it much easier to know where null checks are necessary and where they aren't.
I am not familiar with Nice, but Kotlin implements the same idea, have nullable and not-null types build into the language's type system. A lot more concise than Optionals or null Object pattern. –
Parshall
Common "problem" in Java indeed.
First, my thoughts on this:
I consider that it is bad to "eat" something when NULL was passed where NULL isn't a valid value. If you're not exiting the method with some sort of error then it means nothing went wrong in your method which is not true. Then you probably return null in this case, and in the receiving method you again check for null, and it never ends, and you end up with "if != null", etc..
So, IMHO, null must be a critical error that prevents further execution (that is, where null is not a valid value).
The way I solve this problem is this:
First, I follow this convention:
- All public methods / API always check their arguments for null
- All private methods do not check for null since they are controlled methods (just let die with nullpointer exception in case it wasn't handled above)
- The only other methods which do not check for null are utility methods. They are public, but if you call them for some reason, you know what parameters you pass. This is like trying to boil water in the kettle without providing water...
And finally, in the code, the first line of the public method goes like this:
ValidationUtils.getNullValidator().addParam(plans, "plans").addParam(persons, "persons").validate();
Note that addParam() returns self, so that you can add more parameters to check.
Method validate() will throw checked ValidationException if any of the parameters is null (checked or unchecked is more a design/taste issue, but my ValidationException is checked).
void validate() throws ValidationException;
The message will contain the following text if, for example, "plans" is null:
"Illegal argument value null is encountered for parameter [plans]"
As you can see, the second value in the addParam() method (string) is needed for the user message, because you cannot easily detect passed-in variable name, even with reflection (not the subject of this post anyway...).
And yes, we know that beyond this line we will no longer encounter a null value so we just safely invoke methods on those objects.
This way, the code is clean, easily maintainable, and readable.
Absolutely. Applications that just throw the error and crash are of higher quality because there is no doubt when they are not working. Applications that swallow the errors at best degrade gracefully but usually don't work in ways that are difficult to notice and don't get fixed. And when the problem is noticed they are much harder to debug. –
Accredit
Asking that question points out that you may be interested in error handling strategies. How and where to handle errors is a pervasive architectural question. There are several ways to do this.
My favorite: allow the Exceptions to ripple through - catch them at the 'main loop' or in some other function with the appropriate responsibilities. Checking for error conditions and handling them appropriately can be seen as a specialized responsibility.
Sure do have a look at Aspect Oriented Programming, too - they have neat ways to insert if( o == null ) handleNull() into your bytecode.
In addition to using assert you can use the following:
if (someobject == null) {
// Handle null here then move on.
}
This is slightly better than:
if (someobject != null) {
.....
.....
.....
}
Mh, why that? Please don't feel any defensive, I'd just like to learn more about Java :) –
Proportional
@Mudu As a general rule, I prefer the expression in an if statement to be a more "positive" statement, rather than a "negative" one. So if I saw
if (!something) { x(); } else { y(); } I would be inclined to refactor it as if (something) { y(); } else { x(); } (though one could argue that != null is the more positive option...). But more importantly, the important part of the code is not wrapped inside {}s and you have one level less of indentation for most of the method. I don't know if that was fastcodejava's reasoning but that would be mine. –
Heighten This is what I tend to do as well.. Keeps the code clean in my opinon. –
Woeful
@MatrixFrog Yeah that makes the code cleaner and easily readable, as you see the important code (when everything function without errors) first. –
Axiomatic
Just don't ever use null. Don't allow it.
In my classes, most fields and local variables have non-null default values, and I add contract statements (always-on asserts) everywhere in the code to make sure this is being enforced (since it's more succinct, and more expressive than letting it come up as an NPE and then having to resolve the line number, etc.).
Once I adopted this practice, I noticed that the problems seemed to fix themselves. You'd catch things much earlier in the development process just by accident and realize you had a weak spot.. and more importantly.. it helps encapsulate different modules' concerns, different modules can 'trust' each other, and no more littering the code with if = null else constructs!
This is defensive programming and results in much cleaner code in the long run. Always sanitize the data, e.g. here by enforcing rigid standards, and the problems go away.
class C {
private final MyType mustBeSet;
public C(MyType mything) {
mustBeSet=Contract.notNull(mything);
}
private String name = "<unknown>";
public void setName(String s) {
name = Contract.notNull(s);
}
}
class Contract {
public static <T> T notNull(T t) { if (t == null) { throw new ContractException("argument must be non-null"); return t; }
}
The contracts are like mini-unit tests which are always running, even in production, and when things fail, you know why, rather than a random NPE you have to somehow figure out.
why would this be downvoted? in my experience, this is far superior to the other approaches, would love to know why not –
Baseboard
I agree, this approach prevents problems related to nulls, rather than fixing them by speckling code null checks everywhere. –
Casern
The problem with this approach is that if name is never set, it has the value "<unknown>", which behaves like a set value. Now let's say I need to check if name was never set (unknown), I have to do a string comparison against the special value "<unknown>". –
Menorrhagia
True good point Steve. What I often do is have that value as a constant, e.g. public static final String UNSET="__unset" ... private String field = UNSET ... then private boolean isSet() { return UNSET.equals(field); } –
Baseboard
IMHO, This is a implementation of Null Object Pattern with a yourself implementation of Optional (Contract). How it behaves on persistence class class? I do not see applicable in that case. –
Firehouse
Guava, a very useful core library by Google, has a nice and useful API to avoid nulls. I find UsingAndAvoidingNullExplained very helpful.
As explained in the wiki:
Optional<T>is a way of replacing a nullable T reference with a non-null value. An Optional may either contain a non-null T reference (in which case we say the reference is "present"), or it may contain nothing (in which case we say the reference is "absent"). It is never said to "contain null."
Usage:
Optional<Integer> possible = Optional.of(5);
possible.isPresent(); // returns true
possible.get(); // returns 5
get() is never a good idea to use. You can use: val = funcThatReturnsOptional().orElse(CUSTOM_ERROR_DIFFERENT_USECASES); in one line and check the value for different usecases. –
Nsf
This is a very common problem for every Java developer. So there is official support in Java 8 to address these issues without cluttered code.
Java 8 has introduced java.util.Optional<T>. It is a container that may or may not hold a non-null value. Java 8 has given a safer way to handle an object whose value may be null in some of the cases. It is inspired from the ideas of Haskell and Scala.
In a nutshell, the Optional class includes methods to explicitly deal with the cases where a value is present or absent. However, the advantage compared to null references is that the Optional<T> class forces you to think about the case when the value is not present. As a consequence, you can prevent unintended null pointer exceptions.
In above example we have a home service factory that returns a handle to multiple appliances available in the home. But these services may or may not be available/functional; it means it may result in a NullPointerException. Instead of adding a null if condition before using any service, let's wrap it in to Optional<Service>.
WRAPPING TO OPTION<T>
Let's consider a method to get a reference of a service from a factory. Instead of returning the service reference, wrap it with Optional. It lets the API user know that the returned service may or may not available/functional, use defensively
public Optional<Service> getRefrigertorControl() {
Service s = new RefrigeratorService();
//...
return Optional.ofNullable(s);
}
As you see Optional.ofNullable() provides an easy way to get the reference wrapped. There are another ways to get the reference of Optional, either Optional.empty() & Optional.of(). One for returning an empty object instead of retuning null and the other to wrap a non-nullable object, respectively.
SO HOW EXACTLY IT HELPS TO AVOID A NULL CHECK?
Once you have wrapped a reference object, Optional provides many useful methods to invoke methods on a wrapped reference without NPE.
Optional ref = homeServices.getRefrigertorControl();
ref.ifPresent(HomeServices::switchItOn);
Optional.ifPresent invokes the given Consumer with a reference if it is a non-null value. Otherwise, it does nothing.
@FunctionalInterface
public interface Consumer<T>
Represents an operation that accepts a single input argument and returns no result. Unlike most other functional interfaces, Consumer is expected to operate via side-effects.
It is so clean and easy to understand. In the above code example, HomeService.switchOn(Service) gets invoked if the Optional holding reference is non-null.
We use the ternary operator very often for checking null condition and return an alternative value or default value. Optional provides another way to handle the same condition without checking null. Optional.orElse(defaultObj) returns defaultObj if the Optional has a null value. Let's use this in our sample code:
public static Optional<HomeServices> get() {
service = Optional.of(service.orElse(new HomeServices()));
return service;
}
Now HomeServices.get() does same thing, but in a better way. It checks whether the service is already initialized of not. If it is then return the same or create a new New service. Optional<T>.orElse(T) helps to return a default value.
Finally, here is our NPE as well as null check-free code:
import java.util.Optional;
public class HomeServices {
private static final int NOW = 0;
private static Optional<HomeServices> service;
public static Optional<HomeServices> get() {
service = Optional.of(service.orElse(new HomeServices()));
return service;
}
public Optional<Service> getRefrigertorControl() {
Service s = new RefrigeratorService();
//...
return Optional.ofNullable(s);
}
public static void main(String[] args) {
/* Get Home Services handle */
Optional<HomeServices> homeServices = HomeServices.get();
if(homeServices != null) {
Optional<Service> refrigertorControl = homeServices.get().getRefrigertorControl();
refrigertorControl.ifPresent(HomeServices::switchItOn);
}
}
public static void switchItOn(Service s){
//...
}
}
The complete post is NPE as well as Null check-free code … Really?.
There's a null check in above code -
if(homeServices != null) { which can be changed to homeServices.ifPresent(h -> //action); –
Padauk I like articles from Nat Pryce. Here are the links:
In the articles there is also a link to a Git repository for a Java Maybe Type which I find interesting, but I don't think it alone could decrease the checking code bloat. After doing some research on the Internet, I think != null code bloat could be decreased mainly by careful design.
Michael Feathers has written an short and interesthing text about approaches like the one you mentioned: manuelp.newsblur.com/site/424 –
Mcdade
If there's anything I appreciated most about this answer, it's the "careful design" tip. Whether it boils down to writing code, fixing bugs or checking for nullity, it's DAMN CRUCIAL to make an overall good design and organisation in code, to avoid many redundancies that we still have to deal with today because of bad design choices made a long time ago... –
Axiomatic
I've tried the NullObjectPattern but for me is not always the best way to go. There are sometimes when a "no action" is not appropiate.
NullPointerException is a Runtime exception that means it's developers fault and with enough experience it tells you exactly where is the error.
Now to the answer:
Try to make all your attributes and its accessors as private as possible or avoid to expose them to the clients at all. You can have the argument values in the constructor of course, but by reducing the scope you don't let the client class pass an invalid value. If you need to modify the values, you can always create a new object. You check the values in the constructor only once and in the rest of the methods you can be almost sure that the values are not null.
Of course, experience is the better way to understand and apply this suggestion.
Byte!
Probably the best alternative for Java 8 or newer is to use the Optional class.
Optional stringToUse = Optional.of("optional is there");
stringToUse.ifPresent(System.out::println);
This is especially handy for long chains of possible null values. Example:
Optional<Integer> i = Optional.ofNullable(wsObject.getFoo())
.map(f -> f.getBar())
.map(b -> b.getBaz())
.map(b -> b.getInt());
Example on how to throw exception on null:
Optional optionalCarNull = Optional.ofNullable(someNull);
optionalCarNull.orElseThrow(IllegalStateException::new);
Java 7 introduced the Objects.requireNonNull method which can be handy when something should be checked for non-nullness. Example:
String lowerVal = Objects.requireNonNull(someVar, "input cannot be null or empty").toLowerCase();
May I answer it more generally!
We usually face this issue when the methods get the parameters in the way we not expected (bad method call is programmer's fault). For example: you expect to get an object, instead you get a null. You expect to get an String with at least one character, instead you get an empty String ...
So there is no difference between:
if(object == null){
//you called my method badly!
}
or
if(str.length() == 0){
//you called my method badly again!
}
They both want to make sure that we received valid parameters, before we do any other functions.
As mentioned in some other answers, to avoid above problems you can follow the Design by contract pattern. Please see http://en.wikipedia.org/wiki/Design_by_contract.
To implement this pattern in java, you can use core java annotations like javax.annotation.NotNull or use more sophisticated libraries like Hibernate Validator.
Just a sample:
getCustomerAccounts(@NotEmpty String customerId,@Size(min = 1) String accountType)
Now you can safely develop the core function of your method without needing to check input parameters, they guard your methods from unexpected parameters.
You can go a step further and make sure that only valid pojos could be created in your application. (sample from hibernate validator site)
public class Car {
@NotNull
private String manufacturer;
@NotNull
@Size(min = 2, max = 14)
private String licensePlate;
@Min(2)
private int seatCount;
// ...
}
javax is, by definition, not "core Java". –
Observer I highly disregard answers that suggest using the null objects in every situation. This pattern may break the contract and bury problems deeper and deeper instead of solving them, not mentioning that used inappropriately will create another pile of boilerplate code that will require future maintenance.
In reality if something returned from a method can be null and the calling code has to make decision upon that, there should an earlier call that ensures the state.
Also keep in mind, that null object pattern will be memory hungry if used without care. For this - the instance of a NullObject should be shared between owners, and not be an unigue instance for each of these.
Also I would not recommend using this pattern where the type is meant to be a primitive type representation - like mathematical entities, that are not scalars: vectors, matrices, complex numbers and POD(Plain Old Data) objects, which are meant to hold state in form of Java built-in types. In the latter case you would end up calling getter methods with arbitrary results. For example what should a NullPerson.getName() method return?
It's worth considering such cases in order to avoid absurd results.
The solution with "hasBackground()" has one drawback - it's not thread-safe. If you need to call two methods instead of one, you need to synchronize the whole sequence in multi-threaded environment. –
Wingard
@Wingard You made a contrived example only to point out that this solution has a drawback. If code is not meant to run in multithreaded application then there is no drawback. I could put at you probably 90% of your code that is not thread safe. We're not speaking here about this aspect of code, we're speaking about design pattern. And multithreading is a topic on its own. –
Caylor
Of course in a single-thread application it's not a problem. I've given that comment because sometimes it is a problem. –
Wingard
@Wingard If you study this topic closer you will find out that Null Object design pattern won't fix the multithreading problems. So it's irrelevant. And to be honest, I've found places where this pattern would fit in nicely, so my original answer is a bit wrong, actually. –
Caylor
- Never initialise variables to null.
- If (1) is not possible, initialise all collections and arrays to empty collections/arrays.
Doing this in your own code and you can avoid != null checks.
Most of the time null checks seem to guard loops over collections or arrays, so just initialise them empty, you won't need any null checks.
// Bad
ArrayList<String> lemmings;
String[] names;
void checkLemmings() {
if (lemmings != null) for(lemming: lemmings) {
// do something
}
}
// Good
ArrayList<String> lemmings = new ArrayList<String>();
String[] names = {};
void checkLemmings() {
for(lemming: lemmings) {
// do something
}
}
There is a tiny overhead in this, but it's worth it for cleaner code and less NullPointerExceptions.
stackoverflow.com/questions/1386275/… –
Servomechanism
+1 This I agree with. You should never return half initialized objects. Jaxb related code and bean code is notiroius for this. It is bad practice. All collections should be initialized, and all objects should exist with (ideally) no null references. Consider a object that has a collection in it. Checking that the object is not null, that the collection is not null and that the collection does not contain null objects is unreasonable and foolish. –
Scheldt
This is the most common error occurred for most of the developers.
We have number of ways to handle this.
Approach 1:
org.apache.commons.lang.Validate //using apache framework
notNull(Object object, String message)
Approach 2:
if(someObject!=null){ // simply checking against null
}
Approach 3:
@isNull @Nullable // using annotation based validation
Approach 4:
// by writing static method and calling it across whereever we needed to check the validation
static <T> T isNull(someObject e){
if(e == null){
throw new NullPointerException();
}
return e;
}
Ad. 4. It is not very useful - when you check if a pointer is null, you probably want to call a method on it. Calling a method on null gives you the same behavior - NullPointerException. –
Wingard
Approach 4 is superfluous and already covered by
Objects.requireNonNull() –
Ulyanovsk Java 8 has introduced a new class Optional in java.util package.
Advantages of Java 8 Optional:
1.) Null checks are not required.
2.) No more NullPointerException at run-time.
3.) We can develop clean and neat APIs.
Optional - A container object which may or may not contain a non-null value. If a value is present, isPresent() will return true and get() will return the value.
For more details find here oracle docs :- https://docs.oracle.com/javase/8/docs/api/java/util/Optional.html
It should be noted, that
Optionals are supposed to be used primarily for return values, as empty Objects to work without consecutive NPE problems, not as a replacement for null checks e. g. in legacy or conventional code. –
Ulyanovsk All in all to avoid statement
if (object != null) {
....
}
since java 7 you can use
Objectsmethods:Objects.isNull(object)
Objects.nonNull(object)
Objects.requireNonNull(object)
Objects.equals(object1, object2)
since java 8 you can use Optional class (when to use)
object.ifPresent(obj -> ...); java 8
object.ifPresentOrElse(obj -> ..., () -> ...); java 9
rely on method contract (JSR 305) and use Find Bugs. Mark your code with annotations
@javax.annotation.Nullableand@javax.annotation.Nonnnul. Also Preconditions are available.Preconditions.checkNotNull(object);
In special cases (for example for Strings and Collections) you can use apache-commons (or Google guava) utility methods:
public static boolean isEmpty(CharSequence cs) //apache CollectionUtils
public static boolean isEmpty(Collection coll) //apache StringUtils
public static boolean isEmpty(Map map) //apache MapUtils
public static boolean isNullOrEmpty(@Nullable String string) //Guava Strings
- When you need to assign default value when null use apache commons lang
public static Object defaultIfNull(Object object, Object defaultValue)
explain very well –
Menhaden
public static <T> T ifNull(T toCheck, T ifNull) {
if (toCheck == null) {
return ifNull;
}
return toCheck;
}
What's wrong with this method, I think @Alkalize just want to give a default value if the it's null, which make sense. –
Naval
There is such a method in Apache commons-lang:
ObjectUtils.defaultIfNull(). There is one more general: ObjectUtils.firstNonNull(), which can be used to implement a degrading strategy: firstNonNull(bestChoice, secondBest, thirdBest, fallBack); –
Adaxial You can use FindBugs. They also have an Eclipse plugin) that helps you find duplicate null checks (among other things), but keep in mind that sometimes you should opt for defensive programming. There is also Contracts for Java which may be helpful.
Java 8 has introduced a new class Optional in java.util package. It is used to represent a value is present or absent. The main advantage of this new construct is that No more too many null checks and NullPointerException. It avoids any runtime NullPointerExceptions and supports us in developing clean and neat Java APIs or Applications. Like Collections and arrays, it is also a Container to hold at most one value.
Below are some useful link you can follow
You can avoid most a lot to avoid NullPointerException by just following most of the others answers to the Question, I just want to add few more ways which have been introduced in Java 9 to handle this scenario gracefully and also showcase a few of the older ones can also be used and thus reducing your efforts.
public static boolean isNull(Object obj)Returns true if the provided reference is null otherwise returns false.
Since Java 1.8
public static boolean nonNull(Object obj)Returns true if the provided reference is non-null otherwise returns false.
Since Java 1.8
public static <T> T requireNonNullElse(T obj, T defaultObj)Returns the first argument if it is non-null and otherwise returns the non-null second argument.
Since Java 9
public static <T> T requireNonNullElseGet(T obj, Supplier<? extends T> supplier)Returns the first argument if it is non-null and otherwise returns the non-null value of supplier.get().
Since Java 9
public static <T> T requireNonNull(T obj, Supplier<String> messageSupplier)Checks that the specified object reference is not null and throws a customized NullPointerException otherwise.
Since Java 1.8
Further details about the above functions can be found here.
you can see there are a lot of answers with many upvotes, what is special about your answer? –
Albaalbacete
I just wanted to add a few more functions made available with
Java 9. –
Worldbeater then emphasize on it with a bold font at the beginning of your post –
Albaalbacete
I follow below guidelines to avoid null checks.
Avoid lazy initialization of member variables as much as possible. Initialize the variables in declaration itself. This will handle NullPointerExceptions.
Decide on mutability of member variables early in the cycle. Use language constructs like
finalkeyword effectively.If you know that augments for method won't be changed, declare them as
final.Limit the mutation of data as much as possible. Some variables can be created in a constructor and can never be changed. Remove public setter methods unless they are really required.
E.g. Assume that one class in your application (
A.java) is maintaining a collection likeHashMap. Don't providepublicgetter method in A.java and allowB.javato directly add an element inMap. Instead provide an API inA.java, which adds an element into collection.// Avoid a.getMap().put(key,value) //recommended public void addElement(Object key, Object value){ // Have null checks for both key and value here : single place map.put(key,value); }And finally, use
try{} catch{} finally{}blocks at right places effectively.
Since Java 7 the class java.util.Objects exists.
But since Java 8, you can use Objects.isNull(var) and Objects.nonNull(var) methods of Objects class to do the null pointer check.
For example,
String var1 = null;
Date var2 = null;
Long var3 = null;
if(Objects.isNull(var1) && Objects.isNull(var2) && Objects.isNull(var3))
System.out.println("All Null");
else if (Objects.nonNull(var1) && Objects.nonNull(var2) && Objects.nonNull(var3))
System.out.println("All Not Null");
I find Guava Preconditions to be very useful in this case. I don't like leaving nulls to null pointer exception since the only way to understand an NPE is by locating the line number. Line numbers in production version and development version can be different.
Using Guava Preconditions, I can check null parameters and define a meaningful exception message in one line.
For example,
Preconditions.checkNotNull(paramVal, "Method foo received null paramVal");
One more alternative:
The following simple function helps to hide the null-check (I don't know why, but I haven't found it as part of the same common library):
public static <T> boolean isNull(T argument) {
return (argument == null);
}
You could now write
if (!isNull(someobject)) {
someobject.doCalc();
}
which is IMO a better way of expressing != null.
If you're constantly going to negate the return value, wouldn't it be better to simply write a function
isNotNull? Doesn't that more clearly indicate your intent? –
Nosedive @TMN: your are right, ideally I'd like to have both a
isNull and isNotNull method. –
Weisler I fail to see how this is more concise than "someobject != null". –
Mclean
Compile-time and run-time dev tools like findbugs will have a harder time identifying duplicate null checks. Also, you better hope Java inlines that isNull function, –
Meatball
Wherever you pass an array or a Vector, initialise these to empty ones, instead of null. - This way you can avoid lots of checking for null and all is good :)
public class NonNullThing {
Vector vectorField = new Vector();
int[] arrayField = new int[0];
public NonNullThing() {
// etc
}
}
If you are using java8 or later go for the isNull(yourObject) from java.util.Objects.
Example:-
String myObject = null;
Objects.isNull(myObject); //will return true
Usage: The below code returns a non null value (if the name is not null then that value will be returned else the default value will be returned).
final String name = "Jobin";
String nonNullValue = Optional.ofNullable(name).orElse("DefaultName");
This is not much diferent from
myObject == null and was introduced in Java8 for Predicates in the lambda feature. –
Misgiving Objects.isNull() is available since Java8 (not Java7) –
Tophus
@Tophus Thankd for pointing it out, I have updated the answer, Actually
Objects is introduced in java7, isNull added to it later. –
Grettagreuze This is a poor usage example for
Objects.isNull. The filter here is superfluous. Optional.ofNullable() will return Optional.empty() if name is null, so you could just do: String nonNullValue = Optional.ofNullable(name).orElse("DefaultName"); –
Ozonize In Java 8 you can use type T for local-variable/field/method-argument/method-return-type if it never assigned null (and do not check for null) or type Optional<T> if it can be null. Then use method map for processing T -> and method flatMap for processing T -> Optional<R>:
class SomeService {
@Inject
private CompanyDao companyDao;
// return Optional<String>
public Optional<String> selectCeoCityByCompanyId0(int companyId) {
return companyDao.selectById(companyId)
.map(Company::getCeo)
.flatMap(Person::getHomeAddress)
.flatMap(Address::getCity);
}
// return String + default value
public String selectCeoCityByCompanyId1(int companyId) {
return companyDao.selectById(companyId)
.map(Company::getCeo)
.flatMap(Person::getHomeAddress)
.flatMap(Address::getCity)
.orElse("UNKNOWN");
}
// return String + exception
public String selectCeoCityByCompanyId2(int companyId) throws NoSuchElementException {
return companyDao.selectById(companyId)
.map(Company::getCeo)
.flatMap(Person::getHomeAddress)
.flatMap(Address::getCity)
.orElseThrow(NoSuchElementException::new);
}
}
interface CompanyDao {
// real situation: no company for such id -> use Optional<Company>
Optional<Company> selectById(int id);
}
class Company {
// company always has ceo -> use Person
Person ceo;
public Person getCeo() {return ceo;}
}
class Person {
// person always has name -> use String
String firstName;
// person can be without address -> use Optional<Address>
Optional<Address> homeAddress = Optional.empty();
public String getFirstName() {return firstName;}
public Optional<Address> getHomeAddress() {return homeAddress;}
}
class Address {
// address always contains country -> use String
String country;
// city field is optional -> use Optional<String>
Optional<String> city = Optional.empty();
String getCountry() {return country;}
Optional<String> getCity() {return city;}
}
Another alternative to the != null check is (if you can't get rid of it design-wise):
Optional.ofNullable(someobject).ifPresent(someobject -> someobject.doCalc());
or
Optional.ofNullable(someobject).ifPresent(SomeClass::doCalc);
With SomeClass being someobject's type.
You can't get a return value back from doCalc() though, so only useful for void methods.
You can get the value by using
map combined with get or any of the other methods for getting the value in the Optional, instead of ifPresent. –
Gaffrigged You can use an interceptor before the method call. That is what aspect-oriented programming focus on.
Suppose M1(Object test) is a method and M2 is a method where we apply an aspect before a method call, M2(Object test2). If test2 != null then call M1, otherwise do another thing. It works for all methods with whom you want to apply an aspect for. If you want to apply an aspect for an instance field and constructor you can use AspectJ. Spring can also be the best choice for a method aspect.
are you proposing intercept each method of entire app? –
Firehouse
public class Null {
public static void main(String[] args) {
String str1 = null;
String str2 = "";
if(isNullOrEmpty(str1))
System.out.println("First string is null or empty.");
else
System.out.println("First string is not null or empty.");
if(isNullOrEmpty(str2))
System.out.println("Second string is null or empty.");
else
System.out.println("Second string is not null or empty.");
}
public static boolean isNullOrEmpty(String str) {
if(str != null && !str.isEmpty())
return false;
return true;
}
}
Output
str1 is null or empty.
str2 is null or empty.
In the above program, we've two strings str1 and str2. str1 contains null value and str2 is an empty string.
We've also created a function isNullOrEmpty() which checks, as the name suggests, whether the string is null or empty. It checks it using a null check using != null and isEmpty() method of string.
In plain terms, if a string isn't a null and isEmpty() returns false, it's not either null or empty. Else, it is.
However, the above program doesn't return empty if a string contains only whitespace characters (spaces). Technically, isEmpty() sees it contains spaces and returns false. For string with spaces, we use the string method trim() to trim out all the leading and trailing whitespace characters.
There has a good way to check the null value from JDK. It is Optional.java that has a sea of methods to resolve these problems. Such as follow:
/**
* Returns an {@code Optional} describing the specified value, if non-null,
* otherwise returns an empty {@code Optional}.
*
* @param <T> the class of the value
* @param value the possibly-null value to describe
* @return an {@code Optional} with a present value if the specified value
* is non-null, otherwise an empty {@code Optional}
*/
public static <T> Optional<T> ofNullable(T value) {
return value == null ? empty() : of(value);
}
/**
* Return {@code true} if there is a value present, otherwise {@code false}.
*
* @return {@code true} if there is a value present, otherwise {@code false}
*/
public boolean isPresent() {
return value != null;
}
/**
* If a value is present, invoke the specified consumer with the value,
* otherwise do nothing.
*
* @param consumer block to be executed if a value is present
* @throws NullPointerException if value is present and {@code consumer} is
* null
*/
public void ifPresent(Consumer<? super T> consumer) {
if (value != null)
consumer.accept(value);
}
It is really, really useful to help javer.
With Java 8, you could pass a supplier to a helper method like below,
if(CommonUtil.resolve(()-> a.b().c()).isPresent()) {
}
Above replaces boiler plate code like below,
if(a!=null && a.b()!=null && a.b().c()!=null) {
}
//CommonUtil.java
public static <T> Optional<T> resolve(Supplier<T> resolver) {
try {
T result = resolver.get();
return Optional.ofNullable(result);
} catch (NullPointerException var2) {
return Optional.empty();
}
}
At first blush, this looks like a great answer to the problem I'm having, but it suffers from the design-flaw of "don't use exceptions to control behavior". Essentially, it only works because it relies on the 'resolve(..)" method intentionally generating an NPE, catching it, and then dealing with it to alter behavior. That's just bad. Might still use it, but looking for a better way. –
Dispart
Java 8 now has Optional class that wraps the object in consideration and if a value is present, isPresent() will return true and get() will return the value.
http://www.oracle.com/technetwork/articles/java/java8-optional-2175753.html
OK, I now this has been technically answered a million times but I have to say this because this is an un-ending discussion with Java programmers.
Sorry but I disagree will almost all of above. The reason we have to be testing for null in Java is because must Java programmers don’t know how to handle memory.
I say this because I have a long experience programming in C++ and we don’t do this. In other words, you don’t need to. And note that, in Java, if you hit a dangling pointer you get a normal exception; in C++ this exception normally is not caught and terminates the program.
Don’t want to do this? Then follow some simple rules ala C/C++.
Don’t instantiate things so easily, think that every "new" can get you in lots of trouble and FOLLOW these simple rules.
A class shall access memory in only 3 ways ->
It can "HAVE" class members, and they will follow these rules:
- ALL "HAS" members are created "new" in the constructor.
- You will close /de allocate in destructor or equivalent close() function in Java for that same class and in NO other.
This means that you need to have in mind (just like Java does) who is the owner or parent of each resource and respect that ownership. An object is only deleted by the class who created it. Also ->
Some members will be "USED" but not own or "HAVE". This are "OWN" in another class and passed as arguments to the constructor. Since these are owned by another class, we will NEVER delete or close this, only the parent can.
A method in a class can also instantiate local objects for internal use which will NEVER pass out side of the class, or they should have been normal "has" objects.
Finally for all this to work, you need to have a disciplined design with classes in hierarchy form and making no cycles.
Under this design, AND following the above rules, there is no way that a child class in a hierarchy design will ever access a pointer which was destroyed, because that means that a parent was destroyed before a child, which the hierarchical acyclic design will not allow it.
Finally, also remember when starting your system you should build from top to bottom of the hierarchy and destroy bottom to top. You will never have a null pointer anywhere, or someone is violating the rules.
Agreed, albeit only conceptually. Yeah, one should HAVE A PLAN. But then you need programmers who CAN MAKE PLANS and have the ability to think logically and develop and maintain consistent patterns. And then there are employers whose main focus is money and wages ;-). And then start having a look at the JPA spec to get an impression of what constitutes a consistent pattern/plan and what it requires to develop and/or document it. –
Mclean
This answer is relevant for C++ but not for a garbage collected language like Java. –
Bestial
I respectfully disagree. Good memory handling in JAVA ( or almost any other language) will give you many benefits. From better stability as nullpointers reduced, better performance as lower memory footprint, to better design as relationships are better controlled. I will suggest that the sloppy memory management of many JAVA applications leds to frequent NPEs and memory leaks. –
Occlude
I most emphatically agree with what you wrote, and think it very unfortunate that Java makes no distinction between references that encapsulate ownership and those which do not. Those who would argue "that's what GC is for" miss a critical point: GC avoids the need for immutable objects to have owners, and it also eliminates the possibility of a dangling reference to a deleted object turning into a reference to some other arbitrary object (as can happen in C++). It's hard to write correct code, however, if more than one object encapsulates a mutable object's state as part of its own. –
Lovage
My preferred way of thinking of things is to say that references can encapsulate identity, mutable state, both, or neither. Additionally, references can encapsulate mutable state either because they identify an immutable object, or because the object they identify will never be exposed to anything that might mutate it. It's too bad Java has no concept of such distinctions, since things like equality checking and cloning could be handled 99% automatically if it did. –
Lovage
I believe what you suggest will only address a small subset of null pointer problems in Java, and will not solve the original problem described here. –
Rocketeer
It might be more relevant than you think. More info: ibm.com/developerworks/library/j-jtp06243 :) –
Occlude
You can also use the Checker Framework (with JDK 7 and beyond) to statically check for null values. This might solve a lot of problems, but requires running an extra tool that currently only works with OpenJDK AFAIK. https://checkerframework.org/
The way to avoid unnecessary null-checks is simple to state:
You need to know which variables can be null, and which cannot, and you need to be confident about which category a given variable fall into.
But, although it can be stated simply enough, achieving it is harder. The key lies in the confident part, because how can you be sure that a variable can't be null?
There are no quick-fix, easy answers to this, but here are some pointers:
Clean code. The most important thing for being able to reason about the behaviour of a piece of code is that it is written in a matter that is easy to understand. Name your variables based on what they represent, name your methods after what they do, apply the
Single responsibility principle(theSinSOLID: http://en.wikipedia.org/wiki/SOLID_(object-oriented_design), it means that each piece of code should have a single responsibility, and do this and nothing else). Once your code is clean, it is much easier to reason about it, also across multiple tiers/layers of code. With messy code, trying to understand what a method does might make you forget why you are reading the method in the first place. (Tip: Read "Clean Code" by Robert C. Martin)Avoid returning
nullvalues. If anullvalue would keep your program from functioning correctly, throw anexceptioninstead (make sure to add the appropriate error-handling.) Cases where returning anullvalue might be acceptable is for instance trying to fetch an object from the database. In these cases, write code that handles thenullvalues, and make a note behind your ear that here we have something that might returnnull. Handle returnednullvalues as close to the caller of the method returningnullas possible (don't just blindly pass it back up the call-chain.)Never EVER pass explicit
nullvalues as parameters (at least not across classes). If you are ever in a position where passing anull-parameter is the only option, creating a new method that does not have this parameter is the way to go.Validate your input! Identify the "entry-points" to your application. They can everything from webservices, REST-services, remote EJB classes, controllers, etc. For each method in these entry-points, ask yourself: "Will this method execute correctly if this parameter is null?" If the answer is no, add
Validate.notNull(someParam, "Can't function when someParam is null!");. This will throw anIllegalArgumentExceptionif the required parameter is missing. The good thing about this type of validation in the entry-points, is that you can then easily assume in the code being executed from the entry-point, that this variable will never be null! Also, if this fails, being at the entry-point, debugging is made a lot easier than it would if you just got aNullPointerExceptiondeep down in your code, since a failure like this can only mean one thing: The client didn't send you all the required information. In most cases you want to validate all input parameters, if you find yourself in a position where you need to allow a lot ofnull-values, it might be a sign of a badly designed interface, which needs refactoring/additions to suite the needs of the clients.When working with
Collections, return an empty one rather than null!When working with a database, utilize
not null-constraints. In that way, you'll know that a value read from the database cannot be null, and you won't have to check for it.Structure your code and stick with it. Doing this allows you to make assumptions about the behaviour of the code, for instance if all input to your application is validated, then you can assume that these values will never be null.
If you are not already doing it, write automated tests of your code. By writing tests, you will reason about your code, and you will also become more confident that it does what it's supposed to. Also, automated tests guards you from blunders during refactoring, by letting you know immediatly that this piece of code is not doing what it used to.
You still have to null-check of course, but it can trimmed down to the bare minimum (i.e. the situation where know you might be getting a null-value, instead of everywhere just to be sure.) When it comes to null-checks, i actually prefer to use the ternary operator (but use with care, when you start nesting them they come really messy.)
public String nullSafeToString(final Object o) {
return o != null ? o.toString() : "null";
}
First of all, we can't really remove all null conditions. We can reduce them using @NotNull and @Nullable annotations (as mentioned already). But this needs to be backed by some framework. This is where OVal can help.
The basic idea is object/parameters/constructor should always satisfy preconditions. You can have a whole lot of preconditions such as Nullable, NotNull and OVal would take care that an object should be in a consistent state when invoked.
I guess OVal internally uses AspectJ to validate the preconditions.
@Guarded
public class BusinessObject
{
public BusinessObject(@NotNull String name)
{
this.name = name;
}
...
}
For example,
// Throws a ConstraintsViolatedException because parameter name is null
BusinessObject bo = new BusinessObject(null);
Personally, I would either go with jim-nelson's answer or if I do find a null check is convenient for a specific context I would incorporate lombok into my project and use the @NonNull annotation.
Example:
import lombok.NonNull;
public class NonNullExample extends Something {
private String name;
public NonNullExample(@NonNull Person person) {
super("Hello");
this.name = person.getName();
}
}
Even the @NonNull preface mention:
- or: How I learned to stop worrying and love the NullPointerException.
Good answer. So many answers here do one or more of these unhelpful things: 1. Introduce more boilerplate than the original null-check 2. hide the problem and possibly make bugs worse than an exception 3. tell you to just get good and not write bugs 4. tell you not use any nulls –
Vinegarish
Objects.isNull(null)
If you are using Java8 then you can try this code.
Try using below code if you are not using Java8
Object ob=null;
if(ob==null){ **do something}
Kotlin with null safety is elegant alternative, but it means a larger change.
One Option you have
Use checker framework's @RequiresNonNull on methods. for ex you get this if you call a method annotated as such, with a null argument. It will fail during compile, even before your code runs! since at runtime it will be NullPointerException
@RequiresNonNull(value = { "#1" }) static void check( Boolean x) { if (x) System.out.println("true"); else System.out.println("false"); } public static void main(String[] args) { check(null); }
gets
[ERROR] found : null
[ERROR] required: @Initialized @NonNull Boolean
[ERROR] -> [Help 1]
There are other methods like Use Java 8's Optional, Guava Annotations, Null Object pattern etc. Does not matter as long as you obtain your goal of avoiding !=null
It is possible to define util methods which handles nested null-checks in an almost pretty way with Java 8 lambdas.
void example() {
Entry entry = new Entry();
// This is the same as H-MANs solution
Person person = getNullsafe(entry, e -> e.getPerson());
// Get object in several steps
String givenName = getNullsafe(entry, e -> e.getPerson(), p -> p.getName(), n -> n.getGivenName());
// Call void methods
doNullsafe(entry, e -> e.getPerson(), p -> p.getName(), n -> n.nameIt());
}
/** Return result of call to f1 with o1 if it is non-null, otherwise return null. */
public static <R, T1> R getNullsafe(T1 o1, Function<T1, R> f1) {
if (o1 != null) return f1.apply(o1);
return null;
}
public static <R, T0, T1> R getNullsafe(T0 o0, Function<T0, T1> f1, Function<T1, R> f2) {
return getNullsafe(getNullsafe(o0, f1), f2);
}
public static <R, T0, T1, T2> R getNullsafe(T0 o0, Function<T0, T1> f1, Function<T1, T2> f2, Function<T2, R> f3) {
return getNullsafe(getNullsafe(o0, f1, f2), f3);
}
/** Call consumer f1 with o1 if it is non-null, otherwise do nothing. */
public static <T1> void doNullsafe(T1 o1, Consumer<T1> f1) {
if (o1 != null) f1.accept(o1);
}
public static <T0, T1> void doNullsafe(T0 o0, Function<T0, T1> f1, Consumer<T1> f2) {
doNullsafe(getNullsafe(o0, f1), f2);
}
public static <T0, T1, T2> void doNullsafe(T0 o0, Function<T0, T1> f1, Function<T1, T2> f2, Consumer<T2> f3) {
doNullsafe(getNullsafe(o0, f1, f2), f3);
}
class Entry {
Person getPerson() { return null; }
}
class Person {
Name getName() { return null; }
}
class Name {
void nameIt() {}
String getGivenName() { return null; }
}
(This answer was first posted here.)
We have been using Apache libraries (Apache Commons) for this issue.
ObjectUtils.equals(object, null)
or
CollectionUtils.isEmpty(myCollection);
or
StringUtils.isEmpty("string");
I like the previous answer before, as a practice, of providing initial default values or empty sets for collections to minimize the need.
These can be simple uses that keep you from having NullPointerException or using an empty collection. This doesnt answer the question for what to do with the null object, but these provide some checks for basic validations of the object or collection.
Hope this helps.
I don't like
CollectionUtils.isEmpty and StringUtils.isEmpty as they also do a null check, which is not implied by the method name. –
Menorrhagia I prefer this
public void simpleFunc(SomeObject someObject){
someObject = someObject != null ? someObject : new SomeObject(null);
someObject.doSomething();
}
Of course in my example SomeObject handles gracefully a null parameter. For example logging such event and doing nothing more.
I would recommend immutable objects. This practice is recommend in many coding practices and useful as industries such as financial services.
With immutability, Immutable objects cannot be changed after creation. This means fields are set at construction and never modified. This naturally eliminates nulls for those fields.
public final class Person {
private final String name;
private final int age;
public Person(String name, int age) {
this.name = name;
this.age = age;
}
public String getName() { return name; }
public int getAge() { return age; }
}
Benefits:
- Predictability: Once created, you're guaranteed the object's state won't change unexpectedly.
- Thread safety: Immutable objects are inherently thread-safe.
- Reduced Null Checks: No need to check if fields may have been changed to null.
You can achieve the above solution using code generators like Lombok also.
It can be handled by this:
import org.apache.commons.lang.ObjectUtils;
if(!ObjectUtils.equals(x, null)){
}
how is that better than
x != null? I don't get the java mindset at times. –
Vinegarish The ObjectUtils.equals method is generally used for comparing two objects for equality, and it handles null values gracefully. –
Tropism
yeah but you are not comparing arbitrary objects, one of the objects is literal
null. And x != null handles that gracefully and is very readable. And even if I do want to compare arbitrary objects in a null-safe way, I use the stdlib Objects.equals nowadays, not some dependency. –
Vinegarish This is a big question ans the only valid answer is: You cannot avoid to check for null in java.
The issue comes from the language syntax itself, an object can always be both typed or null. Even an Optional object can be null.
All of the other proposals (annotations, optional ...) add just more boilerplate code around the issue but never really solve it. And even worse will add more and more complexity and corner cases to handle manually.
The only viable solution is to not write the boilerplate yourself and let another tool do it for you. That is when you use Kotlin. null safety was a major concern when they designed the language.
Also the java syntax will never change so deeply because it will imply some major breaking changes, creating a new language in fact (what kotlin already did at the end).
That is the major reason why so many companies moved to kotlin recently. Check some articles about Facebook, Google android ...
And one of the major issue in the java world is that many people learned java so long ago that they are emotionally attached to it and don't want to change their habit. So they try to justify their weird and obsolete pattern instead of just using a meta language, like TypeScript for JavaScript in the frontend world.
The best way to avoid Null Checks in Java, is to properly handle and use exceptions. Null Checks in my experience have become more common and required as you move closer to the front-end, because it's closer to the user who may supply invalid information through the UI (such as, no value, being submitted for a field).
One may argue that you should be able to control what the UI is doing, lest you forget most UI is done through a third party library of some kind, which for example, may return either NULL or an Empty String for a blank text box, depending on the situation or the library.
You can combine the two like this:
try
{
myvar = get_user_supplied_value();
if (myvar == null || myvar.length() == 0) { alert_the_user_somehow(); return; };
process_user_input(myvar);
} catch (Exception ex) {
handle_exception(ex);
}
Another approach people take is to say:
if (myvar && myvar.length() > 0) { };
You could also throw an exception (which is what I prefer)
if (myvar == null || myvar.length() == 0) {
throw new Exception("You must supply a name!");
};
But that's up to you.
Is myvar a boolean or a String in this case? I'm having trouble figuring out what you mean by
!myvar if myvar.length() is present. Is that perhaps the way null is checked for in other languages? –
Vergne @Vergne !myvar should resolve to (myvar != true), and since NULL is never equal to anything, if it is NULL, it will resolve correctly and !myvar should catch it. I believe this is viable in most languages. Basically I'm checking to make sure it's not null, if it is, that short-circuits the secondary check for length. If it is not null, then it checks the length of the variable. In this case it would be a string. More the overall null check vs. exception, I didn't pay too much attention on the language. –
Dogcatcher
Ah, I see. In Java, attempting to use a
! operator on anything other than a primitive boolean (or a Boolean object with auto-unboxing) or an expression would result in a syntax error and wouldn't compile, so I was confused. Thanks for the clarification. –
Vergne @Vergne Interesting! Thank you, now I'm realizing since the answer was for Java, I should have taken more care to stick to Java syntax. Sorry! –
Dogcatcher
You can make one generic Method for object and string so that you can use it through out in your application- This could help you and your colleagues : Create a class eg. StringUtilities and add the method eg. getNullString
public static String getNullString(Object someobject)
{
if(null==someobject )
return null;
else if(someobject.getClass().isInstance("") &&
(((String)someobject).trim().equalsIgnoreCase("null")||
((String)someobject).trim().equalsIgnoreCase("")))
return null;
else if(someobject.getClass().isInstance(""))
return (String)someobject;
else
return someobject.toString().trim();
}
And simply call this method as,
if (StringUtilities.getNullString(someobject) != null)
{
//Do something
}
Functional approach may help to wrap the repetitive null checks and execute anonymous code like the below sample.
BiConsumer<Object, Consumer<Object>> consumeIfPresent = (s,f) ->{
if(s!=null) {
f.accept(s);
}
};
consumeIfPresent.accept(null, (s)-> System.out.println(s) );
consumeIfPresent.accept("test", (s)-> System.out.println(s));
BiFunction<Object, Function<Object,Object>,Object> executeIfPresent = (a,b) ->{
if(a!=null) {
return b.apply(a);
}
return null;
};
executeIfPresent.apply(null, (s)-> {System.out.println(s);return s;} );
executeIfPresent.apply("test", (s)-> {System.out.println(s);return s;} );
You can couple your Class with Unit Testing using a framework like JUnit. This way your code will be clean (no useless checkings) and you will be sure your instances wont be null.
This is one good reason (of many) to use Unit Testing.
For utility classes, you can check that parameters are not null.
In all other cases, you may not have to. Use encapsulation as much as possible, thus reducing the places you feel tempted to check for null.
We can use 'Objects.requireNonNull' to avoid explicit null checks.
For Ex.
public class Main {
private String myString;
public Main(String myString) {
this.myString = Objects.requireNonNull(myString, "myString must not be null");
}
}
This Answer repeats the accepted Answer without adding any value. –
Insectivorous
Null object pattern can be used as a solution for this problem. For that, the class of the someObject should be modified.
public abstract class SomeObject {
public abstract boolean isNil();
}
public class NullObject extends SomeObject {
@Override
public boolean isNil() {
return true;
}
}
public class RealObject extends SomeObject {
@Override
public boolean isNil() {
return false;
}
}
Now istead of checking,
if (someobject != null) {
someobject.doCalc();
}
We can use,
if (!someObject.isNil()) {
someobject.doCalc();
}
Reference : https://www.tutorialspoint.com/design_pattern/null_object_pattern.htm
You are just modifying the check. No real benefit is received. Even though we put extra efforts to implement the suggested pattern, the code is still not clean. –
Thanksgiving
If you implement the Null Object Pattern, instead of returning null, you should return the NullObject. This way, you don't need the check. Just call
object.doCalc(). If the object is the NullObject instance, the doCalc() won't do anything. It will be just an empty implementation of the method. –
Vernacularism You know, you're basically replacing Null, with your own version of Null, then saying use this version of Null, so you don't have to use the other one that came with the language :) –
Dogcatcher
If you are going to use the null-object pattern with inheritance like this, do not make an
isNil method, but use instanceof with a cast (in modern java that should be an implicit smart-cast). isNil does not tell you anything that the type does not. –
Vinegarish Another suggestion is to program defensively - where your classes/functions provide default values that are known and safe, and where null is reserved for true errors/exceptions.
For example, instead of functions that return Strings returning null when there is a problem (say converting a number to a string), have them return an empty String (""). You still have to test the return value before proceeding, but there would be no special cases for exceptions. An additional benefit of this style of programming is that your program will be able to differentiate and respond accordingly between normal operations and exceptions.
-1. That's even worse. Now instead of crashing if you forget to test, your program silently propagates an incorrect value. –
Virus
while I generally think this is bad I would say it is okay-ish for the situation where you are an application over an existing database where both
null and "" mean "missing value" –
Vinegarish © 2022 - 2024 — McMap. All rights reserved.
finalkeyword and you'll live in a even better world. – BruellOptional.ofNullable(x).ifPresent(xV -> xV.someAction());– Ictinus