Java: difference between String a = null and String a = new String()

Question

It's a pretty naive question to ask but I got this doubt while programming for Android. I use to initialize my new strings like String a=null, so in my code there remains a probability to get NullPointerException if I forget to initialize the variable before accessing it but what about the other case. Also which one should be used?

Answer 1

First let's clarify something: You mention that after assigning null to the variable you could forget to initialize it, but by assigning null to it you are in effect initializing it.

public static void main (String args[]){
    String s;       
    System.out.println(s); // compiler error variable may not be initialized
}

vs

public static void main (String args[]){
    String s=null;      
    System.out.println(s); // no compiler error
    System.out.println(s.equals("helo")); // but this will generate an exception
}

So after you do String s=null; there's is no way that you could forget to initialize because you did initialize it.

That being clear, I would recommend you to use a "smart default". In your case perhaps the empty string "" would be a good default value if you want to avoid NullPointerException. In the other hand, sometimes it is desirable that the program produce an exception because it indicates something wrong happened under the hood that should not have happened.

Answer 2

In general you want to keep declaration and initialisation as close as possible to minimise exactly the type of problem you're talking about.

There is also the issue of redundant initialisation where the value null you're assigning is never used which is extra code that harms readability even if the redundant assignment is optimised away by the compiler.

Sometimes assigning some sort of default value is unavoidable, for example if you declare before a try catch, initialise inside and use it afterwards. For other types you can often find a more natural default value such as an empty list.

Answer 3

There are two different views of "initialisation". One is that a variable has a defined value after initialisation. One is that a variable has a useful value. Neither setting this variable to null or to new String () will set it to a useful value.

Using the variable is a bug, whether it is uninitalised or just not set to something useful. The compiler is very good at finding situations where a variable is used while uninitialised (and doesn't allow it). But by setting the variable to null or new String (), you remove the compiler's ability to tell you when there is no useful value. For example:

String a = null;
if (condition1) a = "Yes";
else if (condition2) a = "No";

If neither condition is true, a isn't set to anything useful, but the compiler can't tell you about it anymore. So we have a bug, but you destroyed the compilers ability to report it. If you hadn't initialised a, the compiler would tell you.