Is there any reason that the glu tesselator takes doubles instead of floats? Almost everyone I know uses floats for vertex information, so why use doubles and force 99.5% of people to copy their data to doubles, tesselate it, and copy back to floats? Inaccuracy issues? Does anyone have any insights on this?
You can definitly do tesselation with floats. The accuracy is not a problem most of the time.
If you look at the other glu functions, most ask for doubles, so it is consistent.
Also, remember that glu isn’t exactly designed for performance applications.