Hi Semyon, Here are the changes that you think unnecessary have been reverted back. Other changes I have retained as is unless any user has any comments on the behaviour. So as of now I have retained the same behaviour as the native one.
New web rev: http://cr.openjdk.java.net/~sveerabhadra/8194135/webrev.02/ <http://cr.openjdk.java.net/~sveerabhadra/8194135/webrev.02/> Thanks and regards, Shashi > On 18-Jan-2018, at 9:55 PM, Semyon Sadetsky <[email protected]> > wrote: > > On 01/18/2018 12:31 AM, Shashidhara Veerabhadraiah wrote: >> Hi Semyon, The test cases which were under the HidingSelectionTest are >> copied to MultiSelectionTest along with other the multiple selection test >> cases. On Linux platforms, a text selection is always visible and vanishes >> the moment a second selection is made. Since the Linux specific behavior >> can't be added to the generic HidingSelectionTest, those test cases are >> added along with the original test cases from HidingSelectionTest to the new >> MultiSelectionTest class. > What is the reason to copy HidingSelectionTest to MultiSelectionTest to run > it only on Linux? I don't see this as part requested in the TCK red issue. > > As for selection vanishing only when another text is selected in the same > window. I'm not sure that this should be the rule. At first not all native > apps follow that on Linux. And the second, we need to keep balance between > native behavior and being cross-platform. When the behavior is ambiguous for > the user we have to follow the native way but when there no such issues the > behavior should be cross-platform. Anyway, in my understanding this topic is > out of the scope of the issue you need to fix in this bug. > > --Semyon > >> >> Thanks and regards, >> Shashi >> >> -----Original Message----- >> From: Semyon Sadetsky >> Sent: Wednesday, January 17, 2018 10:58 PM >> To: shashidhara veerabhadraiah <[email protected]> >> Cc: [email protected] >> Subject: Re: [10] JDK-8194135: The content in textArea can not be pasted >> after clicking "Copy" button. >> >> Hi Shashi, >> >> Why you disabled HidingSelectionTest.java for Linux platform? How is it >> related to the TCK test failure? >> >> --Semyon >> >> >> On 01/16/2018 10:27 AM, shashidhara veerabhadraiah wrote: >>> Hi Semyon, I ran those tests and found to be passed on the selected >>> platforms. Since in linux platforms, we can’t have multiple selections >>> visible at the same time, I have made a different test MultiSelectionTest >>> along with other test cases as is in the HidingSelectionTest. This test >>> passes on the linux platform and fails on other platforms because of the >>> behaviour change on the linux platform. >>> >>> Thanks and regards, >>> Shashi >>> >>>> On 16-Jan-2018, at 10:02 PM, Semyon Sadetsky <[email protected]> >>>> wrote: >>>> >>>> Hi Shashi, >>>> >>>> HidingSelectionTest may not pass after your change because the component >>>> selection is not cleared after the input focus transfer within the >>>> containing window. >>>> >>>> --Semyon >>>> >>>> >>>> On 01/13/2018 03:32 AM, shashidhara veerabhadraiah wrote: >>>>> Hi Semyon, I think that the ownsSelection flag handles the cases well and >>>>> no need for any other conditions I felt. I confirmed this by running the >>>>> older test(HidingSelectionTest) on all the platforms and the tests >>>>> passed. Please let me know any particular reasons for adding those >>>>> conditions. >>>>> >>>>> Thanks and regards, >>>>> Shashi >>>>> >>>>>> On 12-Jan-2018, at 10:08 PM, Semyon Sadetsky >>>>>> <[email protected]> wrote: >>>>>> >>>>>> Hi Shashi, >>>>>> >>>>>> Why did change generic DefaultCaret class while only platform specific >>>>>> behavior should be fixed? >>>>>> >>>>>> --Semyon >>>>>> >>>>>> >>>>>> On 01/12/2018 02:22 AM, shashidhara veerabhadraiah wrote: >>>>>>> Hi All, Please review a fix for the bug: >>>>>>> https://bugs.openjdk.java.net/browse/JDK-8194135 >>>>>>> >>>>>>> Webrev: >>>>>>> http://cr.openjdk.java.net/~sveerabhadra/8194135/webrev.00/ >>>>>>> <http://cr.openjdk.java.net/%7Esveerabhadra/8194135/webrev.00/> >>>>>>> >>>>>>> Summary: The text selection visibility behaves differently on >>>>>>> linux/solaris platforms compared to windows or Mac platforms. There >>>>>>> were 2 related fixes that went in this area. There are: >>>>>>> https://bugs.openjdk.java.net/browse/JDK-5100950 and >>>>>>> https://bugs.openjdk.java.net/browse/JDK-8188081. While fixing the bug >>>>>>> 5100950, a mistake was made to clear out the selection data. This >>>>>>> behaviour is NOT the native behaviour as found in other native >>>>>>> applications. The behaviour always is to retain the selection data and >>>>>>> control the visibility of it depending on the conditions. On linux >>>>>>> flavours we display only one selection data and upon toggling active >>>>>>> application we get to see the selection data of that particular >>>>>>> application hence we never loose the data. These behaviour differences >>>>>>> are tested via the test components attached with this fix. >>>>>>> Considering the time, this fix addresses only the awt text components >>>>>>> like textarea and textfield. Both these classes uses the same caret >>>>>>> implementation, hence a change to the XAWTCaret is sufficient to >>>>>>> reflect to both the text components. Swing based text components will >>>>>>> be addressed later if there are any issues found. >>>>>>> >>>>>>> Thanks and regards, >>>>>>> Shashi >
