In Azure canvas radial gradients will behave incorrectly if the inner circle intersects with, or lies outside of the outer circle.

This patch makes Azure use special shaders to create radial gradients.

This fixes some minor bugs.

Fix another small bug.

Upload the right patch

Comment on attachment 544588 [details] [diff] [review] Properly report passing Doesn't this need changes for the new tests that pass?

Comment on attachment 544589 [details] [diff] [review] Fix Azure radial gradient behavior v3 Review of attachment 544589 [details] [diff] [review]: ----------------------------------------------------------------- ::: gfx/2d/DrawTargetD2D.cpp @@ +1408,5 @@ > + rawStops.resize(stops->mStopCollection->GetGradientStopCount()); > + stops->mStopCollection->GetGradientStops(&rawStops.front(), rawStops.size()); > + > + RefPtr<ID3D10Texture1D> tex = CreateGradientTexture(rawStops); > + I'd move the stops getting into the CreateGradientTexture fucntion. @@ +1445,5 @@ > + > + mPrivateData->mEffect->GetVariableByName("DeviceSpaceToUserSpace")-> > + AsMatrix()->SetMatrix(matrix); > + > + float A = pow(dc.x, 2) + pow(dc.y, 2) - pow(dr, 2); I don't know if you care but VS2008 (didn't check 2010) doesn't turn pow(x, 2) into x*x @@ +1454,5 @@ > + mPrivateData->mEffect->GetVariableByName("A")->AsScalar()->SetFloat(A); > + mPrivateData->mEffect->GetTechniqueByName("SampleRadialGradient")-> > + GetPassByIndex(0)->Apply(0); > + } > + } Can this hunk be a separate function? ::: gfx/2d/HelpersD2D.h @@ +172,5 @@ > + } > + > + return false; > +} > + Can we come up with a better name for IsComplexPattern()? ::: gfx/2d/ShadersD2D.fx @@ +126,5 @@ > + Output.PixelCoord.y = ((1.0f - Output.Position.y) / 2.0f) * dimensions.y; > + Output.PixelCoord.xy = mul(float3(Output.PixelCoord.x, Output.PixelCoord.y, 1.0f), DeviceSpaceToUserSpace).xy; > + return Output; > +} > + Can you add some comments about what this shader is doing. @@ +186,5 @@ > + } > + > + return tex.Sample(sSampler, float2(t, 0.5)) * mask.Sample(sMaskSampler, In.MaskTexCoord).a; > +}; > + Some more documentation here wouldn't hurt. It's also worth saying something like: for a more detailed derivation see pixman.

Comment on attachment 544589 [details] [diff] [review] Fix Azure radial gradient behavior v3 This patch is one of the fixes for Azure we need to take in the early mozilla-aurora space.

Updated with review comments. Carrying r+.

I folded these two patches together and pushed them to inbound: http://hg.mozilla.org/integration/mozilla-inbound/rev/c0eaec585ea7

Keeping this open for aurora approval!

Pushed: http://hg.mozilla.org/releases/mozilla-aurora/rev/ce54da315f7a Sadly with the wrong bug number in the comment :(.

The condition returned from IsPatternSupportedByD2D is reversed.

Build:Mozilla/5.0 (Windows NT 6.1; WOW64; rv:7.0) Gecko/20100101 Firefox/7.0 Setting this bug as verified fixed for firefox 7. The fix that landed on Aurora is visible in the Beta repository (i.e. http://hg.mozilla.org/releases/mozilla-beta/file/51f4341b6871/content/canvas/src/nsCanvasRenderingContext2DAzure.cpp). Also there are no failures with the attached tests to check the radial gradient (i.e. test_2d.gradient.radial.cone.top.html)