> There is currently a regression in the gradient painting causing very > strange colors. I checked the last few revisions of the gradient code > and apparently it caused by the recent optimizations. trunk ? branch ? If branch a way to reproduce ? -- Cyrille Berger