A DESCRIPTION OF THE PROBLEM :
Hi, the operator '^' (negation in a character classes) seems not to work.
I provide a source code example where his behavior is totally different in Java 8 and Java 11 


REGRESSION : Last worked in version 8u191

EXPECTED VERSUS ACTUAL BEHAVIOR :
EXPECTED -
Encoding : UTF-8
The output is ooerdqK$Fop22{78ae������������
ACTUAL -
Encoding : UTF-8
The output is ooerdqKFop22{78ae

---------- BEGIN SOURCE ----------
import java.text.Normalizer;
import java.util.regex.Pattern;

/**
 *
 * @author Andres Bel Alonso
 */
public class BugExample {

    /**
     * @param args the command line arguments
     */
    public static void main(String[] args) {
        // need UTF-8 encoding, ensure it
        System.out.println("Encoding : " + System.getProperty("file.encoding"));
        
        // I want to change this input ir order to delete the non ascii characters and non combining diacritical marks
        // but keep ������ and $ 

        String input = "oo����er����������dqK$F����o����p����2������2����{78a������������e������������";
        String str = Normalizer.normalize(input, Normalizer.Form.NFD);
        Pattern pattern = Pattern.compile("[^\\p{ASCII}&&[^\\p{InCombiningDiacriticalMarks}]&&[^������$]]");
        // I make me clean string
        String out = pattern.matcher(str).replaceAll("");
        
        // Java 8 ouput : ooerdqK$Fop22{78ae������������
        // Java 11 ouput : ooerdqKFop22{78ae
        // java 11 output does not complain because it cleans the characters i wanted to keep. Java 8 output is ok
        System.out.println("The output is " + out);
        
        // Finally, using the regex [\\P{ASCII}&&[\\P{InCombiningDiacriticalMarks}]&&[^������$]] works good in java 11
    }
    
}

---------- END SOURCE ----------

FREQUENCY : always