<AWT Dev> RFR: 8243925 Toolkit#getScreenInsets() returns wrong value on HiDPI screens (Windows)

[Sergey Bylokhov]

On 5/14/20 10:52 pm, Prasanta Sadhukhan wrote:
> I believe the test should not be marked headful as no visible component is being shown. Rest looks ok.

If this keyword will be removed then the test will be skipped during our headful testing on headful systems.

> Regards
> Prasanta
> On 14-May-20 7:44 PM, Sergey Bylokhov wrote:
>> On 5/13/20 11:51 pm, Prasanta Sadhukhan wrote:
>>> Shouldn't we need to use ScaleDownY() for top and bottom?
>>> Also, I think you like to rename runPocess() to runProcess() in the test ?
>>
>> Right! I have also updated the test from which I copied the typo:
>> http://cr.openjdk.java.net/~serb/8243925/webrev.05
>>
>>>
>>> Regards
>>> Prasanta
>>> On 08-May-20 11:39 AM, Sergey Bylokhov wrote:
>>>> Hi, Prasanta.
>>>>
>>>> On Linux, it should generally pass because we already scale the insets there,
>>>> but sometimes this test fails due to size mismatches, we have some rounding
>>>> issues during the calculation and conversion from the device to userspace.
>>>>
>>>> It looks like when I commented out the fractional scales I also accidentally
>>>> delete the assertion in the test.
>>>>
>>>> Here is an update test:
>>>> http://cr.openjdk.java.net/~serb/8243925/webrev.04
>>>> The fractional scales were returned, but the test is excluded on Linux. I
>>>> will fix that separately.
>>>>
>>>>
>>>> On 5/7/20 2:27 am, Prasanta Sadhukhan wrote:
>>>>> Hi Sergey,
>>>>>
>>>>> The regression test pass for me on windows and linux even without the fix
>>>>>
>>>>> Regards
>>>>> Prasanta
>>>>> On 30-Apr-20 3:10 AM, Sergey Bylokhov wrote:
>>>>>> Hello.
>>>>>> Please review the fix for jdk/client.
>>>>>>
>>>>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8243925
>>>>>> Fix: http://cr.openjdk.java.net/~serb/8243925/webrev.02
>>>>>>
>>>>>> One more place when we return the data in pixels, without
>>>>>> proper conversation to the user's space.
>>>>>>
>>>>>> On macOS, this conversion is done by the OS, on Linux we already
>>>>>> convert the data. The only missing platform is Windows. Note that
>>>>>> we still have an inadequate specification for this and some other
>>>>>> Toolkit API which states that results are in "in pixels".
>>>>>>
>>>>>> I will file the separate bug to update the spec because this bug
>>>>>> I would like to backport to 11.
>>>>>>
>>>>>> The bug was found during work on
>>>>>> https://bugs.openjdk.java.net/browse/JDK-8231564
>>>>>>
>>>>>>
>>>>
>>>>
>>
>>


-- 
Best regards, Sergey.