[webkit-reviews] review denied: [Bug 66437] [chromium] aliased render surface edges : [Attachment 106162] Patch

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Fri Sep 2 13:58:14 PDT 2011


James Robinson <jamesr at chromium.org> has denied David Reveman
<reveman at chromium.org>'s request for review:
Bug 66437: [chromium] aliased render surface edges
https://bugs.webkit.org/show_bug.cgi?id=66437

Attachment 106162: Patch
https://bugs.webkit.org/attachment.cgi?id=106162&action=review

------- Additional Comments from James Robinson <jamesr at chromium.org>
View in context: https://bugs.webkit.org/attachment.cgi?id=106162&action=review


The uniform stuff seems very bug-prone, there's 4 nearly identical repetitions
in CCRenderSurface.cpp. I also suspect that you're leaking programs.

> Source/WebCore/platform/graphics/FloatQuad.h:121
> +    // Tests whether points are in clock-wise, or counter clock-wise order.

Can you document what this does when all points are colinear?

> Source/WebCore/platform/graphics/chromium/LayerRendererChromium.h:225
> +    OwnPtr<CCRenderSurface::MaskProgramAA> m_renderSurfaceMaskProgramAA;
> +    OwnPtr<CCRenderSurface::ProgramAA> m_renderSurfaceProgramAA;

if you're adding new programs you need to add cleanup code to
LayerRendererChromium::cleanupSharedObjects(). You should hit ASSERTs in debug
mode, have you run the tests in debug?

> Source/WebCore/platform/graphics/chromium/cc/CCRenderSurface.cpp:151
> +		   shaderSamplerLocation =
maskProgramAA->fragmentShader().samplerLocation();
> +		   shaderMaskSamplerLocation =
maskProgramAA->fragmentShader().maskSamplerLocation();
> +		   shaderMatrixLocation =
maskProgramAA->vertexShader().matrixLocation();
> +		   shaderQuadLocation =
maskProgramAA->vertexShader().pointLocation();
> +		   shaderAlphaLocation =
maskProgramAA->fragmentShader().alphaLocation();
> +		   shaderEdgeLocation =
maskProgramAA->fragmentShader().edgeLocation();

this is so horribly repetitive and error-prone. is there any better way to do
this?


More information about the webkit-reviews mailing list