[OpenJDK 2D-Dev] RFR: 8204391: Colors with alpha are painted incorrectly on Linux

[Philip Race]

Actually, although the recommendation is not "as strong", it is still
recommended to use the EDT for initialising AWT apps.
You'll find the same in some other tests I think.
In any case it is just a harmless precaution.

-phil.

On 7/27/18, 3:11 AM, Prasanta Sadhukhan wrote:
> Fix looks good to me.
>
> A nit in the test
>
> 61 SwingUtilities.invokeAndWait(() -> createAndShowGUI());
> Since awt Frame is being created and not JFrame, there is no need of 
> using EDT.
>
> Regards
> Prasanta
> On 7/27/2018 1:12 AM, Phil Race wrote:
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8204931
>> Webrev: http://cr.openjdk.java.net/~prr/8204931/
>>
>> There is a long evaluation in the bug report, so what I write here is 
>> just a summary.
>>
>> This is a regression introduced by 
>> https://bugs.openjdk.java.net/browse/JDK-8176795
>> and specific to the Xrender pipeline
>>
>> The bug is that when using colors with alpha, the alpha is being 
>> ignored and the colour
>> copied rather than blended, since we no longer pass the 
>> "pre-multiplied" flag as false.
>>
>> The fix is to pre-multiply the colour when storing as a pixel even 
>> for an opaque xrender surface.
>>
>> With this fix automated jtreg tests, and tck tests pass, as well as 
>> the manual TCK test that was failing.
>>
>> Additionally the supplied regression test passes on all platforms.
>>
>> -phil
>>
>>
>>
>