<AWT Dev> [11][JDK-8197810]RFR: Test java/awt/Choice/SelectCurrentItemTest/SelectCurrentItemTest.html fails on Windows

Sergey Bylokhov Sergey.Bylokhov at oracle.com
Wed Jun 27 03:56:41 UTC 2018


On 26/06/2018 19:40, Krishna Addepalli wrote:
> Hi Sergey,
> 
> The test with my changes will run on Linux and Mac as well, since basically I'm trying to select a different item, so that the event is always generated.

Yes, an event is generated if different items are selected, but the test 
was created to check a different use-case: when the same item is 
reselected. Before JDK-7171412 an event was expected, but after the fix 
no events should be generated, when the same item reselected. So in the 
test you need to select the same item an check that no events will be 
generated.

> But, we should answer the question for JDK-8014503, so that the behavior is consistent on all platforms.
> 
> Thanks,
> Krishna
> 
> -----Original Message-----
> From: Sergey Bylokhov
> Sent: Wednesday, June 27, 2018 6:05 AM
> To: Krishna Addepalli <krishna.addepalli at oracle.com>; awt-dev at openjdk.java.net
> Subject: Re: <AWT Dev> [11][JDK-8197810]RFR: Test java/awt/Choice/SelectCurrentItemTest/SelectCurrentItemTest.html fails on Windows
> 
> If this was done intentionally then I suggest to revert the expectation of the test. But I think that the test will fail on lin/mac, because one more related bug JDK-8014503 was not fixed.
> So we should run the test on windows only, until JDK-8014503 is not fixed.
> 
> On 26/06/2018 06:58, Krishna Addepalli wrote:
>> Hi Sergey,
>>
>> I think the current behavior is intentionally implemented in awt in Windows. I don't know how JDK-4902933 is marked as resolved, but there is a linked issue JDK-7171412. Here is the link to the patch that was pushed as fix for this issue, and we can see that there is an explicit suppression of event propagation in case of same index being selected again. Pasting the current code for reference:
>> MsgRouting AwtChoice::WmNotify(UINT notifyCode) {
>>       if (notifyCode == CBN_SELCHANGE) {
>>           int selectedIndex = (int)SendMessage(CB_GETCURSEL);
>>
>>           JNIEnv *env = (JNIEnv *)JNU_GetEnv(jvm, JNI_VERSION_1_2);
>>           jobject target = GetTarget(env);
>>           int previousIndex = env->GetIntField(target,
>> selectedIndexID);
>>
>>           if (selectedIndex != CB_ERR && selectedIndex != previousIndex){
>>               DoCallback("handleAction", "(I)V", selectedIndex);
>>           }
>>
>> This is the reason why the test is failing.
>>
>> Hope this clarifies.
>>
>> Thanks,
>> Krishna
>>
>> -----Original Message-----
>> From: Sergey Bylokhov
>> Sent: Tuesday, May 29, 2018 8:48 PM
>> To: Krishna Addepalli <krishna.addepalli at oracle.com>;
>> awt-dev at openjdk.java.net
>> Subject: Re: <AWT Dev> [11][JDK-8197810]RFR: Test
>> java/awt/Choice/SelectCurrentItemTest/SelectCurrentItemTest.html fails
>> on Windows
>>
>> Hi, Krishna.
>> On 08/05/2018 02:59, Krishna Addepalli wrote:
>>> The basic problem is that, the Robot mouse move is moving to position
>>> where item 0 is located (which is already selected), and selecting it.
>>> Since this item is already selected, there is no new item selection
>>> event generated, which is why the test fails.
>> As far as I understand the usecase which your describe was implemented intentionally in this test to verify the bug:
>> https://bugs.openjdk.java.net/browse/JDK-4902933
>>
>> Did you check what is the reason of behavior change?
>>
>>
>> --
>> Best regards, Sergey.
>>
> 
> 
> --
> Best regards, Sergey.
> 


-- 
Best regards, Sergey.


More information about the awt-dev mailing list