[OpenJDK 2D-Dev] CubicCurve2D.solveCubic and containment/intersection bugs.

[Denis Lila]

> From that email on we started diving into the containment methods
> instead of solveCubic and the email you refer to doesn't have a webrev
> link to the code with your changes. Is there a final webrev for the
> solveCubic changes?

I made a webrev, along with 3 regression tests (2 for my previous 2 pushes
and one for solveCubic):
http://icedtea.classpath.org/~dlila/webrevs/cc2d/webrev/

The regression test for solveCubic just tests the equation {0, 0, 1, 1}
for which we used to find only 1 root. I thought about including a
randomized stress test based your trySolve method that you sent a while
ago, but I'm not a big fan of regression tests that have a small chance of
failing even when nothing is wrong. What do you think about this?

Regards,
Denis.

----- Original Message -----

> 
> ...jim
> 
> On 1/21/2011 1:00 PM, Denis Lila wrote:
> >> I think so, but Phil knows more about the scripts that verify a
> >> push.
> >>
> >> I think they do look for the bugids in the regression tests so I
> >> don't
> >> know what happens if they find a mismatch. I think you are probably
> >> OK
> >> as long as one of the tests lists the bug you are claiming to fix
> >> on
> >> that push...
> >
> > By the way, is the cubic solver improvement good to go? I tried to
> > address the last concerns and questions you had about it in this
> > e-mail:
> > http://mail.openjdk.java.net/pipermail/2d-dev/2011-January/001769.html
> > where I also propose some concrete documentation changes.
> >
> > Thank you,
> > Denis.