character subset is not set as expected in JTextField

OS: ALL
Build:
this issue reproduced on JDK 9 141, 6u141b06, 7u131b06, 8u121b07

steps to reproduce:
#Create a new J text field
tfKata = new JTextField(20);
#Set the field only accept a part of unicode characters
getContentPane().add(setPanel(tfHKata,
                                          "Character.UnicodeBlock.KATAKANA"));
tfHKata.addFocusListener(new FocusAdapter() {
                public void focusGained(FocusEvent e) {
                    InputContext ic = tfHKata.getInputContext();
                    Character.Subset[] s = { Character.UnicodeBlock.KATAKANA };
                    ic.setCharacterSubsets(s);
                }
            });

Actual results:
any kind of characters can be typed in all text fields.

Expected Results:
only restricted characters can be typed.