Hi Jim. > Also, I wrote new hit testing code in jdk6 that used bezier recursion > to compute the values and it ran way faster than any root-finding methods > (mainly because all you have to worry about is subdividing enough that > the curve can be classified as above, below, or to the left or right > and you're done), so the containment methods could probably be fixed by > simply using the new code in sun.awt.geom.Curve and this method could > be updated with the new equations you found and left as an "approximate > method" like it always has been?
Well, I already have half the code I need to implement the ideas I wrote, so... why not? However, making solveCubic that good does not really seem to be a high priority, so how about we quickly push just the new implementation I found so we can fix most cases where an obvious root is missed, then push this dashing/AA webrev (which will depend on the new solveCubic), then I can fix the implementation of intersect using the recursive subdivision you mentioned, and then I can take my time finishing the implementation of the ideas from my last e-mail (these bugs/rfes have waited 10+ years - they can wait 1-2 more months). Right now, I would like to give priority to better pisces support of the new parallelogram pipes and this bug: https://bugzilla.redhat.com/show_bug.cgi?id=662230 Here's the webrev for the new solveCubic implementation: http://icedtea.classpath.org/~dlila/webrevs/cc2d/webrev/ Regards, Denis.
