Re: [OpenJDK Rasterizer] Path2D optimizations

2015-03-27 Thread Laurent Bourgès
Jim, Here is a new Path2d patch: http://cr.openjdk.java.net/~lbourges/path2D/Path2D.2/ Changes: - Fix ArrayIndexOutOfBounds in ptCrossing / rectCrossing when coords array is zero-sized (0 capacity constructor): bug present in jdk8 - new tests: Path2dEmpty to test this bug Path2dCopyConstructor to

Re: [OpenJDK Rasterizer] Path2D optimizations

2015-03-07 Thread Laurent Bourgès
Jim, Ok, I will totally rewrite the Path2D test soon. I was busy on marlin optimizing its merge sort variant based on my mapbench's array data (stats...). I will publish soon a new release with an up-to-date report between pisces, marlin & ductus. Cheers, Laurent

Re: [OpenJDK Rasterizer] Path2D optimizations

2015-03-06 Thread Jim Graham
First, the test cases that I asked for were to verify that a path that was cloned was still operational. I see no tests that verify that the returned objects will still function, only tests that examine internal data structures for a metric that may not be appropriate in the future. On 3/6/15