<Swing Dev> [9] Review request for 8065098 JColorChooser no longer supports drag and drop between two JVM instances

[Sergey Bylokhov]

Hi, Alexander.
On 26.11.2014 15:08, Alexander Scherbatiy wrote:
> On 11/24/2014 1:25 PM, Sergey Bylokhov wrote:
>> Hi, Alexander.
>> can you investigate why during this dnd operation 
>> df.isMimeTypeEqual(DataFlavor.javaJVMLocalObjectMimeType is local?
>
>    The TransferHandler.exportAsDrag(...) method uses 
> PropertyTransferable class for any dragged property like color in 
> JColorChooser.
>    PropertyTransferable class always returns 
> "application/x-java-jvm-local-objectref" mime type for a dragged 
> property.
>    It is better to investigate the dragged property mime type changing 
> in a separate issue to not introduce new regressions.
ok. please file separate issue.
The fix looks fine.
>
>>
>> And the test can be improved as well, since the steps are unclear:
>>   1. Compile the java test JColorChooserDnDTest.java:
>>   27             > javac JColorChooserDnDTest.java
>>   28         2. Run the first instance of the java test:
>>   29             > java JColorChooserDnDTest
>> I am sure that the wrong version of the java will be used in this case. 
>     I have updated the test description:
>       http://cr.openjdk.java.net/~alexsch/8065098/webrev.02/
>
>> Is it possible to automate this test?
>
>     The fix is supposed to be backported to the JDK 8u and I am not 
> sure that the automated test which drags a color from one instance
>     of JVM to another will be stable. I would prefer to updated the 
> current test to automated in JDK 9 as part of the fix
>        JDK-8065647 [macosx] Drag and drop does not work for 
> JColorChooser between two JVM instances
>
>    Thanks,
>    Alexandr.
>
>>
>> On 21.11.2014 19:08, Alexander Scherbatiy wrote:
>>> On 11/21/2014 5:41 PM, Alexey Ivanov wrote:
>>>> Hi Alexandr,
>>>>
>>>> Although I'm not a reviewer, the fix look good to me.
>>>>
>>>> The fix works with JColorChooserDnDTest and does not cause 
>>>> ClassCastException in the test for 8024061. Yet the latter test, 
>>>> test/sun/awt/dnd/8024061/bug8024061.java, fails. I propose to 
>>>> remove the additional check that data is null, lines 280-282, from 
>>>> the test.
>>>
>>>     Thank you for the review.
>>>     I have updated the test bug8024061:
>>> http://cr.openjdk.java.net/~alexsch/8065098/webrev.01
>>>
>>>    Thanks,
>>>    Alexandr.
>>>
>>>>
>>>> Regards,
>>>> Alexey
>>>>
>>>> On 21.11.2014 13:38, Alexandr Scherbatiy wrote:
>>>>>
>>>>> Hello,
>>>>>
>>>>> Could you review the fix:
>>>>>   bug: https://bugs.openjdk.java.net/browse/JDK-8065098
>>>>>   webrev: http://cr.openjdk.java.net/~alexsch/8065098/webrev.00
>>>>>
>>>>>   The issue appears after the fix 8024061 Exception thrown when 
>>>>> drag and drop between two components is executed quickly.
>>>>>   In some way the mime type of the dropped color has 
>>>>> javaJVMLocalObjectMimeType on the second JVM.
>>>>>
>>>>>   The issue should be fixed in JDK 8 as the regression so the easy 
>>>>> safe fix that prevents ClassCastException is added.
>>>>>   The full fix that includes a proper synchronization between XAWT 
>>>>> and EDT threads on Linux can be added later.
>>>>>
>>>>> Thanks,
>>>>> Alexandr.
>>>>>
>>>>
>>>
>>
>>
>


-- 
Best regards, Sergey.