[plt-scheme] some more frtime questions

Hi Dave,

You can call the C++ code, but you'll have to make sure not to pass it
any behaviors (or vectors of behaviors).  The easiest way to ensure
this is probably to 'lift' all of your vector operations:

(require (lifted (lib "maths.ss" "fluxus-0.14") vmul vadd vsub ...))

If you do this, you'll also need to lift the constructor and accessors
for your vectors.  Since 'vector' and 'vector-ref' are already bound
in FrTime, the easiest thing might be to use different names, e.g.:

(define (vec3 x y z)
  (lift #f vector x y z))

(define (vec3-x v)
  (lift #t vector-ref v 0))

etc.

Cheers,
Greg

On 2/11/08, Dave Griffiths <dave at pawfal.org> wrote:
> Hi Greg,
>
> >> I've made some more progress with FrTime/Fluxus - I'm working on some
> >> smaller examples to play with collision detection, and it's going well
> >> so far, using the pong example to learn from.
> >
> > Sounds good.  :-)  Note that the pong demo does collision-detection in
> > a pretty naive way, so you'll probably want to find something more
> > clever before long.
> >
> >> I'd just like to confirm that I'll need to write a 3D version of posn -
> >> as I can't seem to use scheme vectors as behaviours. Is this because
> >> FrTime supports structs but not vectors as behaviours?
> >
> > That's strange.  FrTime *should* let you use vectors without any
> > trouble.  Could you send me an example of what you're trying to do and
> > how it doesn't work?
>
> silly me, fixed this - I just wasn't calling value-now on the vector, just
> the elements.
>
> >> Also, if I want to use fully supported 3D vectors this way, I'm assuming
> >> I won't be able to use the vector maths in fluxus - which is part of a
> >> binary extension.
> >
> > I don't understand what this means.  Does fluxus define a custom 3d
> > vector struct and a bunch of operations for it?  If so, there ought to
> > be a way to reuse all that code; it may just require a little extra
> > work on the Scheme and/or FrTime side(s).  That I'd be glad to help
> > with.
>
> fluxus uses ordinary scheme vectors, but defines a bunch of procedures to
> operate on them (implemented in C++ for speed):
> http://www.pawfal.org/Software/fluxus/docs/0.13/en/maths.html
>
> in the following example (which randomly changes the colour of a sphere
> when a cube is collided with it) I've had to make my own 'dist' procedure
> (distance between vectors taken as points) to do the collision calculation
> - if I swap 'dist' for the equivalent in fluxus - 'vdist', it stops
> working.
>
> (require (lib "frisbee.ss" "fluxus-0.14"))
>
> (define player-pos
>   (vector (integral
>               (hold
>                (map-e
>                 (lambda (key)
>                   (case key
>                     ((#\a) -0.01)
>                     ((#\d) 0.01)))
>                 keyboard) 0))
>
>           (+ (integral
>               (hold
>                (map-e
>                 (lambda (key)
>                   (case key
>                     ((#\s) -0.01)
>                     ((#\w) 0.01)))
>                 keyboard) 0)) 5)
>           0))
>
> (define npc-pos (vector 0 0 0))
>
> (define (sq x) (* x x))
>
> (define (dist a b)
>     (sqrt
>      (+ (sq (- (vector-ref a 0) (vector-ref b 0)))
>         (sq (- (vector-ref a 1) (vector-ref b 1)))
>         (sq (- (vector-ref a 2) (vector-ref b 2))))))
>
> (scene
>  (list
>   (cube
>    #:translate player-pos)
>   (sphere
>    #:translate npc-pos
>    #:colour (hold
>              (map-e
>               (lambda (_)
>                 (snapshot/apply
>                  (lambda (a b)
>                    (vector (rndflt) (rndflt) (rndflt)))
>                  player-pos npc-pos))
>               (when-e (< (dist player-pos npc-pos) 2)))
>              (vector 1 1 1))
>    )))
>
>
>