Open Bug 1029670 Opened 11 years ago Updated 2 years ago

SkLinearGradient's shadeSpan() gives different results depending on the start location

Categories

(Core :: Graphics, defect)

x86
macOS
defect

Tracking

()

People

(Reporter: jrmuizel, Unassigned)

References

Details

This happens because we accumulate the error that results from rounding scaleX to get dx. i.e. transform(startx) + n*dx != transform(startx + n) The fix to this is not obvious to me right now so we should probably just fuzz the issue away for now.
Blocks: skia-reftest
I think we could fix this with some careful fixed point math.
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.