RFR: 8217853: Cleanup in the D3D native pipeline [v2]

[Nir Lisker]

On Tue, 2 Aug 2022 07:36:11 GMT, Ambarish Rapte <arapte at openjdk.org> wrote:

>> Yes, in `PsMath.h::computeLight`, there is a check if the light requests attenuation. In general, this is done only for directional lights (that are not attenuated), but in this case we know that this point light is not attenuated so passing 0 skips the redundant calculation. In theory this would improve performance, but because this light can only exist alone I don't expect anything measurable.
>
> I see 4 tests in `PointLightIlluminationTest` fail due to this change, on MacOS and Windows.
> Tests: 
> sphereLowerRightPixelColorShouldBeDarkRed
> sphereUpperRightPixelColorShouldBeDarkRed
> sphereUpperLeftPixelColorShouldBeDarkRed
> sphereLowerLeftPixelColorShouldBeDarkRed
> 
> Error:
> expected:rgba(139,0,0,255) but was:rgba(180,0,0,255)
> 
> The tests pass if isAttenuated is 1.

I'll have a look. What command did you run this test with?

Can you also test the point light attenuation under `tests\system\src\test\java\test\javafx\scene\lighting3D\PointLightAttenuationTest.java`? (I think there's a bit of a mess in the tests folders.)

-------------

PR: https://git.openjdk.org/jfx/pull/789