The reason is that the String literal "abc"
will be turned into a global String instance for all its ocurrences, it will be the same String instance therefore you can be sure that "abc" == "abc"
. It is possible for the compiler to do that because String instances are immutable. However, if you explicitly allocate the String they will be two different instances and they will also be different to the String instance implicitly created by the compiler i.e. new String("abc") != new String("abc")
and "abc" != new String("abc")
.
Another good example to understand what the compiler is doing is to look at this code:
"abc".contains("a");
you see that the literal behaves like an instance of a String type. You may exploit this to minimize programming errors e.g.
// this is OK and the condition will evaluate to false
String myStringValue = null;
if ("abc".equals(myStringValue)) { // false
whereas this code results in NPE:
// this will produce a NPE
String myStringValue = null;
if (myStringValue.equals("abc")) { // NPE