java - Why does the behavior of the Integer constant pool change at 127?

Question

I am not able to understand how the Java Constant Pool for Integer works.

I understand the behavior of Strings, and hence able to justify myself that it is the same case with Integer Constants also.

So, for Integers

Integer i1 = 127;
Integer i2 = 127;
System.out.println(i1==i2); // True

&

Integer i1 = new Integer(127);
Integer i2 = new Integer(127);
System.out.println(i1==i2); // False

Till here everything goes in my head.

What I am not able to digest is, it behaves differently when I increase the integer from 127. This behavior changes after 127, below is the code snippet

Integer i1 = 128;
Integer i2 = 128;
System.out.println(i1==i2); // False. WHY?????

Can somebody help me understand this?

Answer 1

No, the constant pool for numbers doesn't work the same way as for strings. For strings, only compile-time constants are interned - whereas for the wrapper types for integer types, any boxing operation will always use the pool if it's applicable for that value. So for example:

int x = 10;
int y = x + 1;
Integer z = y; // Not a compile-time constant!
Integer constant = 11;
System.out.println(z == constant); // true; reference comparison

The JLS guarantees a small range of pooled values, but implementations can use a wider range if they wish.

Note that although it's not guaranteed, every implementation I've looked at uses Integer.valueOf to perform boxing operations - so you can get the same effect without the language's help:

Integer x = Integer.valueOf(100);
Integer y = Integer.valueOf(100);
System.out.println(x == y); // true

From section 5.1.7 of the JLS:

If the value p being boxed is true, false, a byte, or a char in the range u0000 to u007f, or an int or short number between -128 and 127 (inclusive), then let r1 and r2 be the results of any two boxing conversions of p. It is always the case that r1 == r2.

Ideally, boxing a given primitive value p, would always yield an identical reference. In practice, this may not be feasible using existing implementation techniques. The rules above are a pragmatic compromise. The final clause above requires that certain common values always be boxed into indistinguishable objects. The implementation may cache these, lazily or eagerly. For other values, this formulation disallows any assumptions about the identity of the boxed values on the programmer's part. This would allow (but not require) sharing of some or all of these references.

This ensures that in most common cases, the behavior will be the desired one, without imposing an undue performance penalty, especially on small devices. Less memory-limited implementations might, for example, cache all char and short values, as well as int and long values in the range of -32K to +32K.