The end of section 5.1.7

http://java.sun.com/docs/books/jls/third_edition/html/conversions.html#5.1.7

says:

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

[See also subsequent discussion section]

AFAIK, boxing uses valueOf to achieve this.

On 16 November 2010 13:50, Mikael <mikael-aronsson@telia.com> wrote:
> As far as I know there is no guarantee that valueOf returns the same object
> every time, it just gives class a chance to do it if it wants to.
>
> ----- Original Message ----- From: "lcj.dev" <lcj.dev@gmail.com>
> To: "dev" <dev@harmony.apache.org>
> Sent: Tuesday, November 16, 2010 2:44 PM
> Subject: Question about Character.valueOf()
>
>
>> Hi, all
>> Why this may be failed for other JREs? assertSame(Character.valueOf(c),
>> Character.valueOf(c)).
>> Thanks.
>>
>> please refers to this test case for details:
>> package org.apache.harmony.luni.tests.java.lang;
>>
>> import junit.framework.TestCase;
>>
>> public class CharacterImplTest extends TestCase {
>>
>>   public void test_valueOfC() {
>>       // test the cache range
>>       for (char c = '\u0000'; c < 512; c++) {
>>           Character e = new Character(c);
>>           Character a = Character.valueOf(c);
>>           assertEquals(e, a);
>>
>>           // WARN: this assertion may not be valid on other JREs
>>           assertSame(Character.valueOf(c), Character.valueOf(c));
>>       }
>>       // test the rest of the chars
>>       for (int c = '\u0512'; c <= Character.MAX_VALUE; c++) {
>>           assertEquals(new Character((char) c), Character.valueOf((char)
>> c));
>>       }
>>   }
>> }
>> 2010-11-16
>>
>>
>>
>> lcj.dev
>
>