> > > Should one aim at reproducing the Sun implementation, or at being > > > a 'correct' one? > > > > It is my understanding that we implement the spec; so bugs in Suns > > code are not relevant information for us. > > Thats not totally true, we have some places where we imitate bugs in > SUNs JDK to stay compatible for apps depending on this bug.
Well, I doubt very many apps depend on the bugs I've found so far, they're mostly of the numerical-error type, and those are to be expected to a certain extent when dealing with Beziers. Although the consequences of this one can sometimes be rather dramatic. If anyone would like to see a dramatic example: http://www.qc.physto.se/~sven/ContainsBug.java displays a CubicCurve2D where the entire line y=151 left of the starting point is considered 'inside'. Obviously, someone at Sun forgot the hazards surrounding the special-case of endpoints. (Yes, I'll be submitting an implementation without this bug soon.) /Sven _______________________________________________ Classpath mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/classpath
