<AWT Dev> [10] Review Request: 8186559 [macos] the screen location is wrong when using AWT/SWT bridge with dual monitors

[Semyon Sadetsky]

+1

--Semyon

On 09/18/2017 03:04 PM, Sergey Bylokhov wrote:
> On 9/18/17 09:53, Semyon Sadetsky wrote:
>> On 09/15/2017 02:29 PM, Sergey Bylokhov wrote:
>>
>>> It is broken from the moment when it was implemented(I have 
>>> rechecked on 8u121 which was mentioned in the report), unlike Apple 
>>> jdk where it worked.
>> Then probably the regression label need to be removed from the bug?
>
> The labels is removed.
>
>>
>> Should the main screen also fixed in other places, for example, in 
>> CTrayIcon.m line 137 ?
>
> Yes, it can be changed as well:
> http://cr.openjdk.java.net/~serb/8186559/webrev.01
>
> but unfortunately it will not work properly even after this,
> the new issues created https://bugs.openjdk.java.net/browse/JDK-8187639
>
> The last usage in getTopmostWindowUnderMouseIDImpl seems correct.
>
>>>
>>> On 9/15/17 08:55, Semyon Sadetsky wrote:
>>>> Hi Sergey,
>>>>
>>>> what was caused the regression?
>>>>
>>>> --Semyon
>>>>
>>>> On 9/14/2017 3:46 PM, Sergey Bylokhov wrote:
>>>>> Hello,
>>>>> Please review the fix for jdk10.
>>>>>
>>>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8186559
>>>>> Webrev can be found at: 
>>>>> http://cr.openjdk.java.net/~serb/8186559/webrev.00
>>>>>
>>>>> When our NSView is embedded in native to the SWT-based NSWindow we 
>>>>> incorrectly fetch the location of the NSView. The reason is that 
>>>>> we use wrong screen when we flip the location. Currently we use 
>>>>> [mainScreen] - which "returns the screen object containing the 
>>>>> window with the keyboard focus", but we need the first screen 
>>>>> which contains 0,0 coordinates.
>>>>>
>>>>> In the fix the utility method is reused which internally uses
>>>>> [[NSScreen screens] objectAtIndex:0]
>>>>>
>>>>
>>>
>>>
>>
>
>