[ Previous Section | Next Section | Chapter Index | Main Index ]


Section 8.4

Assertions and Annotations

In this short section, we look briefly at two features of Java that are not covered or used elsewhere in this textbook, assertions and annotations. They are included here for completeness, but they are mostly meant for more advanced programming.

8.4.1  Assertions

Recall that a precondition is a condition that must be true at a certain point in a program, for the execution of the program to continue correctly from that point. In the case where there is a chance that the precondition might not be satisfied -- for example, if it depends on input from the user -- then it's a good idea to insert an if statement to test it. But then the question arises, What should be done if the precondition does not hold? One option is to throw an exception. This will terminate the program, unless the exception is caught and handled elsewhere in the program.

In many cases, of course, instead of using an if statement to test whether a precondition holds, a programmer tries to write the program in a way that will guarantee that the precondition holds. In that case, the test should not be necessary, and the if statement can be avoided. The problem is that programmers are not perfect. In spite of the programmer's intention, the program might contain a bug that screws up the precondition. So maybe it's a good idea to check the precondition after all -- at least during the debugging phase of program development.

Similarly, a postcondition is a condition that is true at a certain point in the program as a consequence of the code that has been executed before that point. Assuming that the code is correctly written, a postcondition is guaranteed to be true, but here again testing whether a desired postcondition is actually true is a way of checking for a bug that might have screwed up the postcondition. This is something that might be desirable during debugging.

The programming languages C and C++ have always had a facility for adding what are called assertions to a program. These assertions take the form "assert(condition)", where condition is a boolean-valued expression. This condition expresses a precondition or postcondition that should hold at that point in the program. When the computer encounters an assertion during the execution of the program, it evaluates the condition. If the condition is false, the program is terminated. Otherwise, the program continues normally. This allows the programmer's belief that the condition is true to be tested; if it is not true, that indicates that the part of the program that preceded the assertion contained a bug. One nice thing about assertions in C and C++ is that they can be "turned off" at compile time. That is, if the program is compiled in one way, then the assertions are included in the compiled code. If the program is compiled in another way, the assertions are not included. During debugging, the first type of compilation is used, with assertions turned on. The release version of the program is compiled with assertions turned off. The release version will be more efficient, because the computer won't have to evaluate all the assertions.

Although early versions of Java did not have assertions, an assertion facility similar to the one in C/C++ has been available in Java since version 1.4. As with the C/C++ version, Java assertions can be turned on during debugging and turned off during normal execution. In Java, however, assertions are turned on and off at run time rather than at compile time. An assertion in the Java source code is always included in the compiled class file. When the program is run in the normal way, these assertions are ignored; since the condition in the assertion is not evaluated in this case, there is little or no performance penalty for having the assertions in the program. When the program is being debugged, it can be run with assertions enabled, as discussed below, and then the assertions can be a great help in locating and identifying bugs.

An assertion statement in Java takes one of the following two forms:

assert condition ;


assert condition : error-message ;

where condition is a boolean-valued expression and error-message is a string or an expression of type String. The word "assert" is a reserved word in Java, which cannot be used as an identifier. An assertion statement can be used anyplace in Java where a statement is legal.

If a program is run with assertions disabled, an assertion statement is equivalent to an empty statement and has no effect. When assertions are enabled and an assertion statement is encountered in the program, the condition in the assertion is evaluated. If the value is true, the program proceeds normally. If the value of the condition is false, then an exception of type java.lang.AssertionError is thrown, and the program will crash (unless the error is caught by a try statement). If the assert statement includes an error-message, then the error message string becomes the message in the AssertionError.

So, the statement "assert condition : error-message;" is similar to

if ( condition == false )
    throw new AssertionError( error-message );

except that the if statement is executed whenever the program is run, and the assert statement is executed only when the program is run with assertions enabled.

The question is, when to use assertions instead of exceptions? The general rule is to use assertions to test conditions that should definitely be true, if the program is written correctly. Assertions are useful for testing a program to see whether or not it is correct and for finding the errors in an incorrect program. After testing and debugging, when the program is used in the normal way, the assertions in the program will be ignored. However, if a problem turns up later, the assertions are still there in the program to be used to help locate the error. If someone writes to you to say that your program doesn't work when he does such-and-such, you can run the program with assertions enabled, do such-and-such, and hope that the assertions in the program will help you locate the point in the program where it goes wrong.

Consider, for example, the root() method from Subsection 8.3.3 that calculates a root of a quadratic equation. If you believe that your program will always call this method with legal arguments, then it would make sense to write the method using assertions instead of exceptions:

 * Returns the larger of the two roots of the quadratic equation
 * A*x*x + B*x + C = 0, provided it has any roots.  
 * Precondition: A != 0 and B*B - 4*A*C >= 0.
static public double root( double A, double B, double C )  {
   assert A != 0 : "Leading coefficient of quadratic equation cannot be zero.";
   double disc = B*B - 4*A*C;
   assert disc >= 0 : "Discriminant of quadratic equation cannot be negative.";
   return  (-B + Math.sqrt(disc)) / (2*A);

The assertions are not checked when the program is run in the normal way. If you are correct in your belief that the method is never called with illegal arguments, then checking the conditions in the assertions would be unnecessary. If your belief is not correct, the problem should turn up during testing or debugging, when the program is run with the assertions enabled.

If the root() method is part of a software library that you expect other people to use, then the situation is less clear. Oracle's Java documentation advises that assertions should not be used for checking the contract of public methods: If the caller of a method violates the contract by passing illegal parameters, then an exception should be thrown. This will enforce the contract whether or not assertions are enabled. (However, while it's true that Java programmers expect the contract of a method to be enforced with exceptions, there are reasonable arguments for using assertions instead, in some cases.) One might say that assertions are for you, to help you in debugging your code, while exceptions are for people who use your code, to alert them that they are misusing it.

On the other hand, it never hurts to use an assertion to check a postcondition of a method. A postcondition is something that is supposed to be true after the method has executed, and it can be tested with an assert statement at the end of the method. If the postcondition is false, there is a bug in the method itself, and that is something that needs to be found during the development of the method.

To have any effect, assertions must be enabled when the program is run. How to do this depends on what programming environment you are using. (See Section 2.6 for a discussion of programming environments.) In the usual command line environment, assertions are enabled by adding the option -enableassertions to the java command that is used to run the program. For example, if the class that contains the main program is RootFinder, then the command

java -enableassertions RootFinder

will run the program with assertions enabled. The -enableassertions option can be abbreviated to -ea, so the command can alternatively be written as

java -ea RootFinder

In fact, it is possible to enable assertions in just part of a program. An option of the form "-ea:class-name" enables only the assertions in the specified class. Note that there are no spaces between the -ea, the ":", and the name of the class. To enable all the assertions in a package and in its sub-packages, you can use an option of the form "-ea:package-name...". To enable assertions in the "default package" (that is, classes that are not specified to belong to a package, like almost all the classes in this book), use "-ea:...". For example, to run a Java program named "MegaPaint" with assertions enabled for every class in the packages named "paintutils" and "drawing", you would use the command:

java  -ea:paintutils...  -ea:drawing...  MegaPaint

If you are using the Eclipse integrated development environment, you can specify the -ea option by creating a run configuration. Right-click the name of the main program class in the Package Explorer pane, and select "Run As" from the pop-up menu and then "Run..." from the submenu. This will open a dialog box where you can manage run configurations. The name of the project and of the main class will be already be filled in. Click the "Arguments" tab, and enter -ea in the box under "VM Arguments". The contents of this box are added to the java command that is used to run the program. You can enter other options in this box, including more complicated enableassertions options such as -ea:paintutils.... When you click the "Run" button, the options will be applied. Furthermore, they will be applied whenever you run the program, unless you change the run configuration or add a new configuration. Note that it is possible to make two run configurations for the same class, one with assertions enabled and one with assertions disabled.

8.4.2  Annotations

The term "annotation" refers to notes added to or written alongside a main text, to help you understand or appreciate the text. An annotation might be a note that you make to yourself in the margin of a book. It might be a footnote added to an old novel by an editor to explain the historical context of some event. The annotation is metadata or "metatext," that is, text written about the main text rather than as part of the main text itself.

Comments on a program are actually a kind of annotation. Since they are ignored by the compiler, they have no effect on the meaning of the program. They are there to explain that meaning to a human reader. It is possible, of course, for another computer program (not the compiler) to process comments. That's what done in the case of Javadoc comments, which are processed by a program that uses them to create API documentation. But comments are only one type of metadata that might be added to programs.

In Java 5.0, a new feature called annotations was added to the Java language to make it easier to create new kinds of metadata for Java programs. This has made it possible for programmers to devise new ways of annotating programs, and to write programs that can read and use their annotations.

Java annotations have no direct effect on the program that they annotate. But they do have many potential uses. Some annotations are used to make the programmer's intent more explicit. Such annotations might be checked by a compiler to make sure that the code is consistent with the programmer's intention. For example, @Override is a standard annotation that can be used to annotate method definitions. It means that the method is intended to override (that is replace) a method with the same signature that was defined in some superclass. A compiler can check that the superclass method actually exists; if not, it can inform the programmer. An annotation used in this way is an aid to writing correct programs, since the programmer can be warned about a potential error in advance, instead of having to hunt it down later as a bug.

To annotate a method definition with the @Override annotation, simply place it in front of the definition. Syntactically, annotations are modifiers that are used in much the same way as built-in modifiers like "public" and "final." For example,

@Override public void WindowClosed(WindowEvent evt) { ... }

If there is no "WindowClosed(WindowEvent)" method in any superclass, then the compiler can issue an error. In fact, this example is based on a hard-to-find bug that I once introduced when trying to override a method named "windowClosed" with a method that I called "WindowClosed" (with an upper case "W"). If the @Override annotation had existed at that time -- and if I had used it -- the compiler would have rejected my code and saved me the trouble of tracking down the bug.

(Annotations are a fairly advanced feature, and I might not have mentioned them in this textbook, except that the @Override annotation can show up in code generated by Eclipse and other integrated development environments.)

There are two other standard annotations. One is @Deprecated, which can be used to mark deprecated classes, methods, and variables. (A deprecated item is one that is considered to be obsolete, but is still part of the Java language for backwards compatibility for old code.) Use of this annotation would allow a compiler to generate warnings when the deprecated item is used.

The other standard annotation is @SurpressWarnings, which can be used by a compiler to turn off warning messages that would ordinarily be generated when a class or method is compiled. @SuppressWarnings is an example of an annotation that has a parameter. The parameter tells what class of warnings are to be suppressed. For example, when a class or method is annotated with


then no warnings about the use of deprecated items will be emitted when the class or method is compiled. There are other types of warning that can be suppressed; unfortunately the list of warnings and their names is not standardized and will vary from one compiler to another.

Note, by the way, that the syntax for annotation parameters -- especially for an annotation that accepts multiple parameters -- is not the same as the syntax for method parameters. I won't cover the annotation syntax here.

Programmers can define new annotations for use in their code. Such annotations are ignored by standard compilers and programming tools, but it's possible to write programs that can understand the annotations and check for their presence in source code. It is even possible to create annotations that will be retained at run-time and become part of the running program. In that case, a program can check for annotations in the actual compiled code that is being executed, and take actions that depend on the presence of the annotation or the values of its parameters.

Annotations can help programmers to write correct programs. To use an example from the Java documentation, they can help with the creation of "boilerplate" code -- that is, code that has a very standardized format and that can be generated mechanically. Often, boilerplate code is generated based on other code. Doing that by hand is a tedious and error-prone process. A simple example might be code to save certain aspects of a program's state to a file and to restore it later. The code for reading and writing the values of all the relevant state variables is highly repetitious. Instead of writing that code by hand, a programmer could use an annotation to mark the variables that are part of the state that is to be saved. A program could then be used to check for the annotations and generate the save-and-restore code. In fact, it would even be possible to do without that code altogether, if the program checks for the presence of the annotation at run time to decide which variables to save and restore.

[ Previous Section | Next Section | Chapter Index | Main Index ]