<AWT Dev> [9] RFR for 8151714: [TEST] add a manual test test for JOptionPane dialog multires. icons

[Alexander Scherbatiy]

The fix looks good to me.

Thanks,
Alexandr.

On 17/03/16 11:15, Alexander Stepanov wrote:
> Hello Alexandr,
>
> > it should be created and used on EDT
> the following methods affecting UI are run on EDT: UI(), 
> showDialogOrFrame() and dispose() operations in doTest (the final 
> 'dispose' is indeed should be run on EDT as well). The robot 
> operations/checks are executed on main thread using volatile/final 
> references to get the coordinates.
>
> But if this is still not safe enough, then this 'extends JFrame' could 
> be easily removed as it doesn't make any special sense.
>
> Please see the updated fix:
> http://cr.openjdk.java.net/~avstepan/8151714/webrev.02/
> (the transform-based scale check was also removed as it seems that the 
> fix for JDK-8150844 is near to be pushed).
>
> Thanks,
> Alexander
>
>
> On 3/16/2016 8:55 PM, Alexander Scherbatiy wrote:
>>
>> The MultiResolutionJOptionPaneIconTest extends JFrame so it shoud be 
>> created and used on EDT as others Swing components.
>>
>> Thanks,
>> Alexndr.
>>
>> On 14/03/16 18:26, Alexander Stepanov wrote:
>>> Hello Semyon,
>>>
>>> > JOptionPane.showInternalInputDialog() is an utility method
>>> Yes, the initial intention was to check the icons for all the 
>>> dialogs created by means of these utilities. But in principle these 
>>> options could be skipped, and the test could be limited by checking 
>>> the icons for JOptionPane's createDialog() and createInternalFrame() 
>>> methods.
>>>
>>> In such a case the test indeed may be easily converted to the 
>>> automated one.
>>>
>>> > What doesn't work exactly, the dispose() call?
>>> Please disregard. Simply shouldn't create and dispose the dialog at 
>>> the same EDT.
>>>
>>> So please see the updated webrev:
>>> http://cr.openjdk.java.net/~avstepan/8151714/webrev.01/
>>>
>>> Thanks,
>>> Alexander
>>>
>>> On 3/11/2016 8:38 PM, Semyon Sadetsky wrote:
>>>>
>>>>
>>>> On 3/11/2016 7:38 PM, Alexander Stepanov wrote:
>>>>> > just call dispose().
>>>>>
>>>>> 1.
>>>>>         SwingUtilities.invokeAndWait(() -> { 
>>>>> JOptionPane.showInternalInputDialog(...); });
>>>>>         r.waitForIdle();
>>>>>
>>>>> dispose what? no reference to the dialog shown. or do you mean 
>>>>> again use of component tree?
>>>> JOptionPane.showInternalInputDialog() is an utility method.
>>>>>
>>>>> 2.
>>>>>
>>>>> this trivial way:
>>>>>
>>>>>         JOptionPane op = new JOptionPane(msg, info, opt, icon);
>>>>>         JDialog dlg = op.createDialog(parentPane, "test");
>>>>>         SwingUtilities.invokeAndWait(() -> { dlg.setVisible(true); });
>>>>>         r.waitForIdle(2000);
>>>>>         dlg.dispose();
>>>>>
>>>>> - doesn't work as well until "ok" button isn't pressed by the user 
>>>>> (and it differs from the initial test variety). not sure if it 
>>>>> could be fixed by some thread tricks.
>>>> What doesn't work exactly, the dispose() call?
>>>>
>>>> --Semyon
>>>>>
>>>>> Thanks,
>>>>> Alexander
>>>>>
>>>>> On 3/11/2016 7:17 PM, Semyon Sadetsky wrote:
>>>>>>
>>>>>>
>>>>>> On 3/11/2016 6:33 PM, Alexander Stepanov wrote:
>>>>>>> > traverse the component tree
>>>>>>> Hm, interesting. But I'm not sure if everything is smoothly as, 
>>>>>>> e.g., it is also required to close the dialogs one-by-one (for 
>>>>>>> now the user does that), and I'm not sure if it is easy to 
>>>>>>> implement.
>>>>>> Seriously? It is damn easy: just call dispose().
>>>>>>>
>>>>>>> On 3/11/2016 5:55 PM, Semyon Sadetsky wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>> On 3/11/2016 5:45 PM, Alexander Stepanov wrote:
>>>>>>>>> *inside of the internal pane
>>>>>>>>> parent pane
>>>>>>>>>
>>>>>>>>> On 3/11/2016 5:42 PM, Alexander Stepanov wrote:
>>>>>>>>>> Hello Semyon,
>>>>>>>>>>
>>>>>>>>>> I'm not sure if we can control the location of the dialogs 
>>>>>>>>>> shown (and of course we don't know the location of the icons 
>>>>>>>>>> on them), can we?
>>>>>>>> Yes, we can: Component# getLocationOnScreen() and traverse the 
>>>>>>>> component tree to find the Label component with the icon.
>>>>>>>>
>>>>>>>> --Semyon
>>>>>>>>>> It is clear that the internal dialogs are located somewhere 
>>>>>>>>>> inside of the internal pane, but this is not true for the 
>>>>>>>>>> others.
>>>>>>>>>>
>>>>>>>>>> So (probably I'm wrong) there is seemingly no an elegant way 
>>>>>>>>>> to predict the location of the icons, and simple iteration 
>>>>>>>>>> over the screen coordinates (checking for the pixel color) 
>>>>>>>>>> looks not very reliable.
>>>>>>>>>>
>>>>>>>>>> Thanks,
>>>>>>>>>> Alexander
>>>>>>>>>>
>>>>>>>>>> On 3/11/2016 5:24 PM, Semyon Sadetsky wrote:
>>>>>>>>>>> Hi Alexandr,
>>>>>>>>>>>
>>>>>>>>>>> Can the test be automated? It is possible to read color from 
>>>>>>>>>>> the screen with AWT Robot.
>>>>>>>>>>>
>>>>>>>>>>> --Semyon
>>>>>>>>>>>
>>>>>>>>>>> On 3/11/2016 4:53 PM, Alexander Stepanov wrote:
>>>>>>>>>>>> Hello,
>>>>>>>>>>>>
>>>>>>>>>>>> Could you please review the fix
>>>>>>>>>>>> http://cr.openjdk.java.net/~avstepan/8151714/webrev.00/
>>>>>>>>>>>> for
>>>>>>>>>>>> https://bugs.openjdk.java.net/browse/JDK-8151714
>>>>>>>>>>>> - just a single test added.
>>>>>>>>>>>>
>>>>>>>>>>>> Thanks,
>>>>>>>>>>>> Alexander
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/awt-dev/attachments/20160317/b4298eeb/attachment.html>