Re: mesa tesselator

Jyrki Alakuijala wrote:
> Jon Harrop wrote:
>> Have a look at the GLU polygon tesselator.
>
> I have witnessed a project using the GLU polygon tesselator
> that worked fine until the polygons got remarkably larger,
> and somewhere between chiliagons and myriagons the GLU polygon
> tesselator crashes.

We have developed a commercial application which makes extensive use of the
GLU polygon tesselator and find it to be quite robust:

http://www.ffconsultancy.com/products/presenta/

There used to be a bug which caused it to segfault when given a vertex with
an infinite coordinate. I'm not sure if that is still there.

> Checking the versioning history of MESA reveals quite some
> bugs in the tesselator and even a report of "reverting
> back to tesselator 1.1", hinting that building a tesselator
> might not be quite as easy as one would initially expect.

It would be pretty stupid to expect writing a GLU tesselator to be easy. I
tried to write a replacement once and gave up having read three mathematics
PhD theses on the subject.

> If you decide to use the MESA GLU polygon tesselator, you
> may be just fine, but test the code most carefully that it
> really works in the conditions where you really need it to
> work.

I've been using it for years and it definitely really works.

> Sometimes, especially in human safety critical scientific
> computation, such as structural engineering, medical
> computing or nuclear engineering, you do not want to rely too
> much on sloppy things like OpenGL-drivers

Absolutely. But I doubt anyone writing a safety critical application would
rely upon OpenGL. At least, I hope they wouldn't...

--
Dr Jon D Harrop, Flying Frog Consultancy
http://www.ffconsultancy.com
.