Sunday 1 March 2015

Java 7 Main Features

It’s been a while since the last major Java release and expectations were naturally high for the upcoming release. The Java 7 release initially included many JSRs with exciting features, like support for closures, which were later deferred to Java 8 in order to release JSRs that are already done. This effectively diluted what is now offered in Java 7 and has left some disappointed.

Most of the changes were driven by the Java Community Process (JCP) which was established in 1998 as a formal and transparent process to let interested individuals and entities participate and influence how the language should evolve. This is done through the submission of a change request, known as Java Specification Request (JSR), followed by a review and a voting process. Changes or enhancements made to the language can be usually tracked back to a JSR where they were originally put forward for review. For example, the addition of Generics in Java 5 was done via JSR 14.

Java Releases

Here’s a quick snapshot of the past Java release dates (table 1). There are several small new features and enhancements in Java 7. Out of the 28 features that I looked at, here are the ones that I found useful.

New features and enhancements

#1 Strings in switch

In programming, we often encounter situations where we need to do different things based on different values of a variable. For Boolean variables, an if-then-else statement is the perfect way of branching code. For primitive variable types we use the switch statement. However, for String variables, we tend to resort to using multiple if-then-else branches as follows.

Java 6 and Before

One workaround for this is to convert the String into an enum and then switch on the enum.

Java 7

Java 7, however, has added a language level support for String in switch. Now you can rewrite the same code more elegantly:

Not only does this help us write more readable code, but it also helps the compiler generate more efficient byte code as compared to the if-then-else by actually switching on the hashcode() and then doing an equals() comparison. Please note that you will get a NullPointerException if the variable language in the above example resolves to null. I like this feature, but unlike some of the other enhancements in the past (like Generic in Java 5), I don’t anticipate using this feature a lot. Practically, I find myself using if-then-else for one or two values and resort to an Enum when the number of values are higher.

#2 try-with-resources statement

One of the most useful additions in Java 7 is the auto closing of resources like InputStream which helps us reduce boiler plate code from our programs. Suppose we were writing a program which reads a file and closes the FileInputStream when it’s done, here is how you would write the program:

With Java 6 and Before

I want to point out a couple of things in this code. Firstly, notice that we declare the FileInputStream outside the try block just so that it can be accessed in the finally block. The second observation is that we need to initialize the InputStream to null, so that it is guaranteed to be initialized when we access it in the finally block. Last but not the least, the is.close() in the finally block may throw an Exception as well, thereby hiding the original Exception thrown in the try block Exception from the caller. What we probably want is to handle the Exception thrown from is.close() and throw the original IOException.

The above code still has a shortcoming that the Exception thrown from finally is supressed and not accessible to the calling code. I’m not sure how often we want to get both the original Exception and also the Exception thrown from the finally block, but if we did want it, we could do always do something like this:

SuppressedException above is a user written Java bean with a field named suppressed of type Exception. The calling code can then call SupressedException.getThreadLocal().getException() to get the Exception that was supressed in the finally clause. Great, we solved all the problems associated with the try-catch-finally! Now we must remember to repeat this exact sequence with each use of try-catch-finally when handling files or other resources which need to be closed. Enter Java 7, and we can do the above without the boiler plate code.

With Java 7

try can now have multiple statements in the parenthesis and each statement should create an object which implements the new java.lang.AutoClosable interface. The AutoClosable interface consists of just one method.

Each AutoClosable resource created in the try statement will be automatically closed! If an exception is thrown in the try block and another Exception is thrown while closing the resource, the first Exception is the one eventually thrown to the caller. The second Exception is available to the caller via the ex.getSupressed() method. Throwable.getSupressed() is a new method added on Throwable in Java 7 just for this purpose.

Mandatory catch block when using the try statement

Note that if you are creating AutoClosable resources in the try block, you are required to declare a catch block to catch the concrete exception thrown from the actual AutoClosable. close() method. Alternatively you need to throw the Exception. Think of the close() method as implicitly being called as the last line in the try block. So, if an application has its own AutoClosable class as follows:

Then, the following will cause a compile error:

To fix the above, you need to catch or throw the Exception from the calling method.

Syntax for declaring multiple resources

The try statement can contain multiple statements separated by semicolon. Each statement must result in an AutoClosable object.

It is illegal to declare any variable which isn’t an AutoClosable.


ERROR: try-with-resources not applicable to variable type
required: AutoCloseable
found: long

AutoClosable vs Closable

The old Closable interface introduced in Java 5, which also has the same method that now extends from AutoClosable, implying that all Closable classes are automatically Auto-Closable. Those classes automatically become resources that can be created in the try statement. The slight difference in AutoClosable and Closable is that unlike Closable.close(), AutoClosable.close() is not supposed to be idempotent, which means that calling it twice can have side effects. The second difference is that since exceptions thrown from AutoClosable. close() are suppressed, AutoClosable.close() should not throw exceptions which can cause problems when suppressed, like the InterruptedException.

#3 More precise rethrow

There are often situations when we want to catch multiple exceptions of different types, do “something” with them, and rethrow them. Let us consider this example, where we have some code which throws IOException and SQLException.

In the above example, we are logging each type of exception being thrown by the try block before rethrowing them. This results in a duplication of code in the catch blocks. Before Java 7, to get around the code duplication issue we would catch the base exception as follows:

However, this requires us to rethrow the base Exception type java.lang.Exception from the calling method.
public static void main(String[] args) throws Exception {

With Java 7

Java 7 has added the “precise rethrow” feature which lets you catch and throw the base exception while still throwing the precise exception from the calling method.

Note the keyword final in the above catch clause. When a parameter is declared final, the compiler statically knows to only throw those checked exceptions that were thrown by the try block and were not caught in any preceding catch blocks.

#4 Multi-catch

There is no elegant way with Java 6 to catch multiple exceptions of different types and handle them in the same way.

You can always catch the parent Exception in order to avoid duplication of code, but it is not always suitable especially if the parent is java.lang.Exception. Java 7 lets you catch multiple Exceptions in a single catch block by defining a “union” of Exceptions to be caught in the catch clause.

Note that the pipe ‘|’ character is used as the delimiter. The variable ‘ex’ in the above example is statically typed to the base class of Ex1 and Ex2, which is java.lang.Exception in this case.

#5 Binary integral literals

With Java 7, you can now create numerical literals using binary notation using the prefix “0b”

int n = 0b100000;
System.out.println(“n = ” + n);


n = 32

#6 Underscores in numeric literals

With Java 7, you can include underscores in numeric literals to make them more readable. The underscore is only present in the representation of the literal in Java code, and will not show up when you print the value.

Without underscore

int tenMillion = 10000000;
System.out.println(“Amount is “ + tenMillion);



With underscore

int tenMillionButMoreReadable = 10_000_000;
System.out.println(“Amount is ” + tenMillionButMoreReadable);



More rules and examples

1. Consecutive underscores is legal.

int n = 1000______000;

2. Underscores can be included in other numeric types as well.

double d = 1000_000.0d;
long l = 1000_000l;
int hex = 0xdead_c0de;
int bytes = 0x1000_0000;

3. Underscore can be included after the decimal point.

double example8 = 1000_000.000_000d;

4. It is illegal to start a literal with an underscore

5. It is illegal to end a literal with an underscore.

6. It is also illegal to have underscore be just before or after a decimal point.

7. Improved type inference for generic instance creation

Java 5 introduced generics which enabled developers to write type safe collections. However, generics can sometimes be too verbose. Consider the following example where we are creating a Map of List of String.

With Java 5 and 6

Map<String, List<String>> retVal = new HashMap<String, List<String>>();

Note that the full type is specified twice and is therefore redundant. Unfortunately, this was a limitation of Java 5 and 6.

With Java 7

Java 7 tries to get rid of this redundancy by introducing a left to right type inference. You can now rewrite the same statement by using the <> construct.

Map<String, List<String>> retVal = new HashMap<>();

This does make the code a little less verbose. You can also use <> construct when returning a newly created object.

This, in my opinion, goes only half way. The full solution would have been a right to left full type inference.

Map map = new HashMap<String, String>();

The above would have made the code even less verbose. Though this enhancement can still be done in a later version.

#8 More new I/O APIs for the Java platform (NIO.2)

A new set of I/O APIs and features were introduced in Java 1.4 under the java.nio package. This addition was called the New I/O APIs, also known as NIO. Naming something New is always short-sighted because it will not remain new forever. When the next version comes along, what should the new version be called, the NEW NEW I/O? Java 1.7 offers a rich set of new features and I/O capabilities, called NIO.2 (New I/O version 2?). Here are the key highlights of NIO.2.

a) Package

The most important package to look for is java.nio.file. This package contains many practical file utilities, new file I/O related classes and interfaces.

b) The java.nio.file.Path interface

Path is probably the new class that developers will use most often. The file referred by the path does not need to exist. The file referred to does not need to exist. For all practical purposes, you can think of replacing with java. io.Path.

Old Way

File file = new File(“hello.txt”);
System.out.println(“File exists() == ” + file.exists());

New Way

Path path = FileSystems.getDefault().getPath(“hello.txt”);
System.out.println(“Path exists() == ” + Files.exists(path));

c) The java.nio.file.Files class

The Files class offers over 50 utility methods for File related operations which many developers would have wanted to be a part of earlier Java releases. Here are some methods to give you a sense of the range of methods offered. • copy() – copy a file, with options like REPLACE_EXISTING, NOFOLLOW_LINKS public static Path copy(Path source, Path target, CopyOption… options);

move() – move or rename a file public static Path move(Path source, Path target, CopyOption… options);

newInputStream() – create input stream public static InputStream newInputStream(Path path, OpenOption… options);

readAllBytes() – similar to the Apache IOUtils.readFile-ToByteArray public static byte[] readAllBytes(Path path) throws IOException;

createSymbolicLink() – creates a symbolic link, if supported by the file system public static Path createSymbolicLink(Path link, Path target, FileAttribute<?>… attrs) throws IOException

d) WatchService API

WatchService API is a new feature introduced in Java 1.7. It provides an API that lets you “listen” to a certain type of file system events. Your code gets called automatically when those events occur. Examples of event types are captured by StandardWatchEventKinds class.

• ENTRY_CREATE:an entry is created or renamed in the directory
• ENTRY_DELETE:an entry is created or renamed out of the directory
• ENTRY_MODIFY:a directory entry is modified

Here’s a full example of how to watch a directory and print any newly created files.

Run the above program. Then create a file ‘new.txt’ in the directory ‘logs’. The program will print:
logs: new file created new.txt

Note about WatchService implementation

The implementation will take advantage of native support for file change notification when supported by the native file system, but will resort to polling otherwise.

first appeared


Post a Comment


Get in touch with me


Bangalore, India