Re: [Openexr-devel] Request for trivial addition?

[Florian Kainz]

Hi Scott,

sorry for the late reply.  I will add a Sphere::intersectT()
function to Imath (and, in the header file, I will try to 
document what intersect() and intersectT() actually do).  
The changes should show up in the next release of OpenEXR.

Florian


"E. Scott Larsen" wrote:
> 
> As I was flipping through the code this afternoon to get a look at the
> usefulness of it all, I was struck by the <object>::intersect(Line3....)
> functions (specifically in Imath/ImathSphere.h and Imath/ImathPlane.h).
> For the plane, there are two:
> 
> bool Plane::intersect(const Line3<T>&, Vec3<T>&)const;
> bool Plane::intersectT(const Line3<T>&, T&)const;
> 
> the first passes back the point of intersection and the second the
> distance along the line of the intersection.  Good.  For the sphere, there
> is only one:
> 
> bool Sphere::intersect(const Line3<T>&, Vec3<T>&)const;
> 
> which finds the point of intersection.  The distance along the line is
> sure useful though.  Yea, distance from a point to a point will give that,
> but that's costly, and the dumb thing then would be that the distance is
> already computed!  Implementation of Sphere::intersect() looks like this:
> {
> //find the distance along the line of the intersection, call it t
> //get the point via the nice line(t) call
> }
> So, would it be possible to add an additional function:
> 
> bool Sphere::intersectT(const Line3<T>&, T&)const;
> 
> which is implemented exactly like the other one, except it doesn't do the
> last step and passes t back instead of the point?  This is exactly what
> the Plane code does (identical copy, just omits the last line).
> 
> This should take 48 seconds, and I can give you a patch if you want, any
> format you like (whoever you is).
> 
> Thanks!
> 
> //Scott
> 
> _______________________________________________
> Openexr-devel mailing list
> address@hidden
> http://mail.nongnu.org/mailman/listinfo/openexr-devel