[burchard: Fun with incompatible pointers...]

[munzner]

The fantasy (or "documentation", as we say in the trade):

/u/gcg/ngrap/include/pick.h:47
> #define PA_FACE		109	/* Point3 *; array of ...

versus the reality:

/u/gcg/ngrap/src/lib/gprim/geom/pick.c:216
>    case PA_FACE: *(HPoint3 **)attrp = p->f; break;

/u/gcg/ngrap/src/lib/gprim/polylist/plpick.c:68
>      HPt3Transform(T, &pl->p[found].v[i]->pt, &p->f[i]);

Note that in contrast, PA_POINT really _is_ only calculated
as a Point3, as far as I can tell.  This inconsistency
does not appear to serve any useful purpose.

PB