ADDITIONAL SYSTEM INFORMATION :
Mac OS X
OpenJDK Runtime Environment Homebrew (build 11.0.28+0)
OpenJDK 64-Bit Server VM Homebrew (build 11.0.28+0, mixed mode)


A DESCRIPTION OF THE PROBLEM :
Unicode characters u00D8 and u00F8 (Latin Capital/Small Letter O With Stroke) are only different in case (one lowercase, the other uppercase), and so should be considered equal when using Locale.ROOT and collation strength "primary". Similar to 'A' and 'a', for this strength.

STEPS TO FOLLOW TO REPRODUCE THE PROBLEM :
javac -encoding utf-8 Test.java
java Test

EXPECTED VERSUS ACTUAL BEHAVIOR :
EXPECTED -
With Collator.PRIMARY, Locale.ROOT
'O' and 'o' are considered the same
'Ö' and 'ö' are considered the same
'Ø' and 'ø' are considered the same
ACTUAL -
With Collator.PRIMARY, Locale.ROOT
'O' and 'o' are considered the same
'Ö' and 'ö' are considered the same
'Ø' and 'ø' are considered different

---------- BEGIN SOURCE ----------
import java.text.Collator;
import java.util.Locale;
public class Test {
    public static void main(String... args) {
        Collator collator = Collator.getInstance(Locale.ROOT);
        collator.setStrength(Collator.PRIMARY);
        System.out.println("With Collator.PRIMARY, Locale.ROOT");
        System.out.println("'O' and 'o' are considered " +
                (collator.compare("O", "o") == 0 ? "the same" : "different"));
        System.out.println("'Ö' and 'ö' are considered " +
                (collator.compare("Ö", "ö") == 0 ? "the same" : "different"));
        System.out.println("'\u00D8' and '\u00F8' are considered " +
                (collator.compare("\u00D8", "\u00F8") == 0 ? "the same" : "different"));
    }
}
---------- END SOURCE ----------