Hello All,


I am having trouble using the library because some of the tests in the vgl_conic class (for example) are not written to handle floating point precision error.  Some of the math that “should” be zero comes out as 1e-300 or so.  Tests against zero fail, when they should pass.  This is causing the program to go into infinite loops at some points.  My attempt to fix this involved creating a custom type that I could overload operator== for, so that I could instead check that the difference between the two operands was less than some tolerance value.  I am having much trouble getting this stuff to compile with a custom type.  I know that some of these problems have to do with the compiler not handling implicit template creation correctly, but I seem to be traveling down an almost never ending road, to the point where I’m having to explicitly instantiate entire functions.  Any help with this would be great.



Nick Owens