Re: [9] Review request for 8033699: Incorrect radio button behavior

[Vivi An]

Thanks Alexander. Will make the fix.

Regards,

Vivi


On 10/9/2014 2:52 AM, Alexander Zvegintsev wrote:

Hello Vivi

We can move Hashset creation to btnsInGroup's declaration, thus the following lines can be removed

  460             if (btnsInGroup == null)
  461                 btnsInGroup = new HashSet<JRadioButton>();

Otherwise the fix looks good to me.
Thanks,

Alexander.
On 10/09/2014 09:53 AM, Vivi An wrote:

Thank you Alexander,

New patch with fix:
http://cr.openjdk.java.net/~van/8033699/webrev.08/

Regards,

Vivi

On 10/8/2014 11:36 AM, Alexander Zvegintsev wrote:

Hello Vivi,

getFocusTransferBaseComponent():
             Window parentWnd = SwingUtilities.getWindowAncestor((Component) 
activeBtn);
             if (parentWnd instanceof Container) {
                 Container container = (Container)parentWnd;

Window extending Container class, hence there is no need to use instanceof, null check is enough here. container variable is redundant here, you may reuse parentWnd.

There are many redundant class casts in this function.
Also I suggest to use ternary operator to increase readability.

Something like this:

        Component getFocusTransferBaseComponent(boolean next) {
            Component focusBaseComp = activeBtn;

Window container = SwingUtilities.getWindowAncestor(activeBtn);

            if (container != null) {

FocusTraversalPolicy policy = container.getFocusTraversalPolicy();

                Component comp = next
                        ? policy.getComponentAfter(container, activeBtn)

: policy.getComponentBefore(container, activeBtn);

// If next component in the button group, use last/first button as base focus

                // otherwise, use the activeButton as the base focus
                if (containsInGroup(comp)) {
                    focusBaseComp = next ? lastBtn : firstBtn;
                }
            }

            return focusBaseComp;
        }

Small typo in test: Seperate - > Separate (2 times)

--
Thanks,
Alexander.
On 10/08/2014 09:45 AM, Vivi An wrote:

Hello Alexander,

Latest patched is ready for review, test case is updated too:
http://cr.openjdk.java.net/~van/8033699/webrev.07/

Thanks

Vivi

On 10/2/2014 2:46 PM, Vivi An wrote:

Thanks Alexander. Will look into it.

Regards,

~ Vivi

On 10/2/2014 12:20 AM, Alexander Zvegintsev wrote:

Sorry, I wasn't clear,

provided code with the "MIDDLE" button is just an example what user can write (pretty odd one btw)

it is not in your tests.

By lost functionality I mean that after this fix this user can't navigate to the "MIDDLE" button via keyboard

(or to the btnEnd after ButtonGroup reordering).

--
Thanks,
Alexander.
02.10.2014 0:40, Vivi An wrote:

Thanks Alexander.

I have some questions,  please see comments inline.

Regards,

~ Vivi

On 10/1/2014 5:47 AM, Alexander Zvegintsev wrote:

Hello Vivi,

the fix looks good to me in general, but we lost some functionality,

By functionality, you mean the functionality of the test, am I right?

here is some modified code from the previous version of the test:

        ButtonGroup btnGrp = new ButtonGroup();
        btnGrp.add(radioBtn1);
        btnGrp.add(radioBtn2);
        btnGrp.add(radioBtn3);
        radioBtn1.setSelected(true);

        mainFrame.getContentPane().add(btnStart);
        mainFrame.getContentPane().add(radioBtn1);
        mainFrame.getContentPane().add(radioBtn2);
        mainFrame.getContentPane().add(new JButton("MIDDLE"));
        mainFrame.getContentPane().add(radioBtn3);
        mainFrame.getContentPane().add(btnEnd);

Just double checked the test in previous version and current version, did not find a button "MIDDLE" like previous code snippet, or I misunderstood, this is pseudo code?

Compare to the old test code, new changes made are:

1) Radio button group( radioBtn1, radioBtn2 and radioBtn1 ) added to a panel ("box" in below code snippet) and then to main frame, border also added too to separate from new added non grouped single radio button 2) A single radio button (Not Grouped) is added to cover the bug found in last round 3) Tests 1- 6 are adjusted for preceding change to cover areas would like to test

Below is snippet from latest version, the "box" is the panel which contains radio button group, a screenshot added too

  110         mainFrame.getContentPane().add(btnStart);
  111         mainFrame.getContentPane().add(box);
  112         mainFrame.getContentPane().add(radioBtnSingle);
  113         mainFrame.getContentPane().add(btnEnd);

After this fix we are not able to select "MIDDLE" button with a keyboard anymore.

If we change ButtonGroup order to

        btnGrp.add(radioBtn3);
        btnGrp.add(radioBtn2);

We get into a loop and we are unable to select btnEnd and all following components(Shift+Tab can help in that case).

I think that these cases deserves some attention.

--
Thanks,
Alexander.

On 10/01/2014 01:15 AM, Vivi An wrote:

Hello Alexander,

Thanks for the review.

New patch is available under http://cr.openjdk.java.net/~van/8033699/webrev.06/

Regards,

Vivi

On 9/19/2014 3:22 AM, Alexander Zvegintsev wrote:

Hello Vivi,

Since you are touching this code, could you please add missing @Override annotations?

Added

void jumpToNextComponent(boolean next) {
            if (!getButtonGroupInfo())
                return;

This leads to an issue: we can't switch to a next component with a TAB key if JRadioButton is not in a ButtonGroup. (It may be a reason to write another test.)

Fixed

    private boolean isValidRadioButtonObj(Object obj) {

return ((obj != null) && (obj instanceof JRadioButton) &&

                    ((JRadioButton) obj).isVisible() &&
                    ((JRadioButton) obj).isEnabled());
    }

Fixed

There is no need to do a null check before instanceof [1]

     * @param evt, the event object.
     * @param next, indicate if it's next one
     */

private void selectRadioButton(ActionEvent event, boolean next) {

typo: evt should be event

Fixed

I run the test and it fails at least on Linux. It happens due to a big auto delay: Robot holds key for too long so system sends TAB key multiple time.

So I think auto delay should be about 100 ms.

robot.setAutoDelay(300);

It is enough to call it once after Robot creation, this will automatically add a 300 ms delay between Robot generated events (such as keyPress, mouseMove, etc). If you really want a delay between test runs please use Thread.sleep().

I see many calls to hitKey() and then to realSync(), so we can move realSync() call to the end of hitKey(). checkResult() may be combined with runTest(), hence we call one function instead of two.

        int sum = 0;
        for (int i = 0; i < testResults.length; i++) {
            if (testResults[i] == false)
                sum++;
        }

        if (sum != 0){

System.out.println("Total 5 tests, failed test(s): " + sum);

            throw new RuntimeException("Test failed.");
        }

I don't think that this logic is applicable here, each test depends on previous: Test2 failure means that

focus is on a wrong component, so all following tests will fail.

BTW, this code is unreachable in case of test failure, since checkResult() already throws an exception.

http://docs.oracle.com/javase/specs/jls/se7/html/jls-15.html#jls-15.20.2

Test is updated, non-grouped radio button test is added, test passed on Oracle Linux 5 u6 32 bits and Windows 7.

--
Thanks,
Alexander.

On 09/18/2014 09:23 PM, Vivi An wrote:

Thank you Alexandr.

Need one more review for this. Alexander, could you please take a look at below fix?

http://cr.openjdk.java.net/~van/8033699/webrev.05/

Thanks

Vivi

On 9/18/2014 12:55 AM, Alexander Scherbatiy wrote:

  The fix looks good to me.

  Thanks,
  Alexandr.

On 9/18/2014 3:09 AM, Vivi An wrote:

Hello Alex,

On 9/17/2014 12:44 AM, Alexander Scherbatiy wrote:

On 9/17/2014 9:17 AM, Vivi An wrote:

You are right. Fixed now.

http://cr.openjdk.java.net/~van/8033699/webrev.04/

 528             if (e.isShiftDown()) {

529 if (e.getKeyCode() == KeyEvent.VK_TAB) { 537 // ... btnGroupInfo.jumpToNextComponent(false);

 539                 }

540 } else if (e.getKeyCode() == KeyEvent.VK_TAB) { 548 // ... btnGroupInfo.jumpToNextComponent(true);

 550             }

The code in the if/else branches is almost the same except the true/false argument.

  I would suggest to unify them in the following way:

     if (e.getKeyCode() == KeyEvent.VK_TAB) {
        // ... jumpToNextComponent(!e.isShiftDown())
     }

Fixed.

 510             if (next)
 511 KeyboardFocusManager.

512 getCurrentKeyboardFocusManager().focusNextComponent((JComponent)lastBtn);

 513             else
 514 KeyboardFocusManager.

515 getCurrentKeyboardFocusManager().focusPreviousComponent((JComponent)firstBtn);

 This code can be also a little bit optimized:

KeyboardFocusManager.getCurrentKeyboardFocusManager().

focusPreviousComponent((JComponent) (next ? lastBtn : firstBtn));

It's different function call, one for focusNextComponent and the other for focusPreviousComponent, not able to optimize in suggested way.

http://cr.openjdk.java.net/~van/8033699/webrev.05/

Thanks

Vivi

 Thanks,
 Alexandr.

Thank you

~ Vivi

On 9/16/2014 12:39 AM, Alexander Scherbatiy wrote:

  I have missed one more case:
 527         public void keyPressed(KeyEvent e) {
 528             if (e.getKeyCode() == KeyEvent.VK_TAB) {
                        // jumpToNextComponent(true)
 538             }
 539

540 if (e.getKeyCode() == KeyEvent.VK_TAB && e.isShiftDown()) {

//jumpToNextComponent(false)
 550             }
 551         }

It seems that if e.isShiftDown() is true then both jumpToNextComponent(true) and jumpToNextComponent(false) methods can be called.

  Thanks,
  Alexandr.


On 9/16/2014 7:38 AM, Vivi An wrote:

Fixed. New Webrev: http://cr.openjdk.java.net/~van/8033699/webrev.03/

Thanks Alex.

Regards,

~ Vivi

On 9/15/2014 2:59 AM, Alexander Scherbatiy wrote:

On 9/12/2014 8:56 PM, Vivi An wrote:

Thanks Alexander.

All items addressed except last one, please see comments inline. New Webrev: http://cr.openjdk.java.net/~van/8033699/webrev.02/

    163         if ( keyListener != null ) {

Could you remove spaces in the brackets? After code formatting it should be "if (keyListener != null) {"

543 if (next && btnGroupInfo.lastBtn != null)
 544 KeyboardFocusManager.

545 getCurrentKeyboardFocusManager().focusNextComponent((JComponent)btnGroupInfo.lastBtn); 546 else if (btnGroupInfo.firstBtn != null)

 547 KeyboardFocusManager.

548 getCurrentKeyboardFocusManager().focusPreviousComponent((JComponent)btnGroupInfo.firstBtn);

 549         }

Should the first button be focused If next is true and last button is null?

Don't think last button could be null when this function is triggered, last and first will always be set in case there is at least one valid (enabled and visible) radio button in the group.

It seems that lastBtn != null and firstBtn != null checks are not necessary in this case.

  Thanks,
  Alexandr.

Thanks,
Alexandr.

Regards,

~ Vivi


On 9/8/2014 5:43 AM, Alexander Scherbatiy wrote:

On 9/4/2014 10:56 PM, Vivi An wrote:

Hello,

Please review fix for JDK-8033699. Changes made:

1) When tab or shift-tab key pressed, focus moved to next/previous

component outside of the radio button group

2) Using up/down or left/right arrow key to change selection inside

radio button group

Bug:https://bugs.openjdk.java.net/browse/JDK-8033699 Webrev:http://cr.openjdk.java.net/~van/8033699/webrev.00/

  56     private KeyListener keyListener = null;
  57     private KeyHandler handler = null;

It seems that these both fields point to the same object. Is it possible to get rid of one of them?

 152         if ( keyListener != null ) {
 153 button.removeKeyListener( keyListener );
 154         }

Some UIs also set the listener to null in the uninstallUI()/uninstallListeners() methods (like BasicComboBoxUI or BasicColorChooserUI). Should the keyListener also be set to null to free the KeyHandler object?

369 if (!(eventSrc instanceof JRadioButton && 370 ((JRadioButton) eventSrc).isVisible() && ((JRadioButton) eventSrc).isEnabled()))

This check is used several times. It can be moved to a separate method.

373 // Get the button model from the source. 374 ButtonModel model = ((AbstractButton) eventSrc).getModel(); 375 if (!(model instanceof DefaultButtonModel))

 376             return;
 377

378 // If the button model is DefaultButtonModel, and use it, otherwise return. 379 DefaultButtonModel bm = (DefaultButtonModel) model;

 380         if (bm == null)
 381             return;

The second check is not necessary because (model instanceof DefaultButtonModel) returns false for null model.

 404 AbstractButton curElement = null;

The curElement variable declaration could be moved into the 'while' block.

408 if (curElement instanceof JRadioButton &&

 409 ((JRadioButton) curElement).isVisible() &&
 410 ((JRadioButton) curElement).isEnabled()){

It is possible to use 'continue' here to not put the other code inside the 'if' block.

 418 else if (!srcFound){

422 } else if (srcFound && nextBtn == null){

It is not necessary to check the srcFound in the second 'if' because it should already have true value.

 444             if (newSelectedBtn != null){
 445 newSelectedBtn.requestFocusInWindow();
 446 newSelectedBtn.setSelected(true);
 447             }


Is it possible here that newSelectedBtn == eventSrc?

522 private void jumpToNextComponent(JRadioButton btn, boolean next){

The code that retrieves elements from a button group is also used in the selectRadioButton()

and can be moved to a separate method.

Thanks,
Alexandr.

JPRT build succeeded, JCK tests for JRadiobutton and JCheckBox all passed.

Thank you

Regards,

Vivi