Jdeveloper shouldn't suggest not using new + contructor for Strings

Question

Hullo all,
I found a snippet of code on the net which was quite instructive . 5 string creation/assignments and 5 equality tests :
String s1 = "abc";
String s2 = s1;
String s5 = "abc";
String s3 = new String("abc");
String s4 = new String("abc");
System.out.println("== comparison :" + (s1 == s5));
System.out.println("== comparison : " + (s1 == s2));
System.out.println("Using equals method : " + s1.equals(s2));
System.out.println("== comparison : " + s3 == s4);
System.out.println("Using equals method : " + s3.equals(s4));
now, in Jdev11g if you create a String using "new" it grumbles about it : "superfluous string constructor should be removed"
However, it turns out that String s1 = "zzz" and String s1 = new String("zzz") are very different things in java.
Run the code above to see what I mean. There are five comparison tests. You would expect them all to be "true" but
the fourth comparison : System.out.println("== comparison : " + s3 == s4); returns FALSE.
My understanding is that this is due to the optimization of "immutibility of strings" When you say :
String s1 = "zzz" and String s2 = "zzz" then the characters "zzz" exist only once in the heap and both s1 and s2, secretly, reference the same storage, at least until the value is changed.
However using the 'new' keyword forces the JVM to create two separate strings, hence the false result for test #4.
It's nit-picking I know, but I personally found the above code snippet quite instructive, which shows the difference between the statements.

Answer 1

Hi Colin,
good point about the difference in how specifc JVM's (my assumption) can treat this differently.
You can turn off this behaviour via Preferences->Audit->Profiles->Edit then under Rules->API->String Constructor
Best,
John