ScopeTracker

Engine/source/util/scopeTracker.h

Helper class to track the position of a point in N-dimensional space relative to a collection of N-dimensional volumes.

More...

void

_onScopeIn(Object object)

void

_onScopeOut(Object object)

void

_setScope(Object object)

void

_insertTrackingNode(U32 dimension, NodeType * node)

void

_removeTrackingNode(U32 dimension, NodeType * node)

void

_moveTrackingNode(U32 dimension, NodeType * node, F32 newPos)

void

_initTracking()

void

_uninitTracking()

_Anonymous_ {
  MIN = 0
  MAX = 1
}

ObjectType::TrackingNode

NodeType 

TypeTraits< Object >::BaseType

ObjectType 

void

Parent 

NodeType

mBoundaryNodes [NUM_DIMENSIONS][2]

Vector< Object >

mPotentialScopeInObjects 

Object

mReferenceObject 

NodeType *

mTrackingList [NUM_DIMENSIONS][2]

ScopeTracker()

void

debugDump()

Object

getReferenceObject()

void

registerObject(Object object)

void

setReferenceObject(Object object)

void

unregisterObject(Object object)

void

updateObject(Object object)

Detailed Description

Helper class to track the position of a point in N-dimensional space relative to a collection of N-dimensional volumes.

This class works by breaking the N-dimensional case down into N one-dimensional cases. By tracking objects independently along each of their axes, intersection testing becomes a trivial matter of doing point-on-line tests. The line segments can be conveniently represented as ordered linked lists along which the reference point is being moved.

To determine N-dimensional containment of the reference point, the result of each of the one-dimensional point-on-line tests simply has to be combined. If the point lies on each of N 1D segments of a given volume, then the point is fully contained in the volume.

This class may be used in places where otherwise a spatial subdivision scheme would be necessary in order to limit the number of containment tests triggered by each movement of the reference point. Once the tracker has been set up, each position change of an object or the reference center will result in a usually small number of incremental list updates.

Another advantage is that this class makes it easy to reduce 3D tracking to 2D tracking if tracking on the height axis isn't important.

The following interface must be implemented by the given "Object" type:

struct Object : public ScopeTrackerObject< NUM_DIMENSIONS >
{
   /// Return the position of the object in world-space.
   void getPosition( F32 pos[ NUM_DIMENSIONS ] ) const;

   /// If this object is the reference object, this method should return the world-space pivot
   /// point in the object that will be the world reference center.
   void getReferenceCenter( F32 pos[ NUM_DIMENSIONS ] ) const;

   /// Return the object's bounding box in world-space.
   void getBounds( F32 minBounds[ NUM_DIMENSIONS ], F32 maxBounds[ NUM_DIMENSIONS ] ) const;
};

Terminology:

• "In Scope": A volume is in scope if it fully contains the reference center.

• "Reference Object": Object that is the designated center of the world.

Parameters:

_onScopeIn(Object object)

_onScopeOut(Object object)

_setScope(Object object)

_insertTrackingNode(U32 dimension, NodeType * node)

_removeTrackingNode(U32 dimension, NodeType * node)

_moveTrackingNode(U32 dimension, NodeType * node, F32 newPos)

_initTracking()

_uninitTracking()

@218

typedef ObjectType::TrackingNode NodeType 

typedef TypeTraits< Object >::BaseType ObjectType 

typedef void Parent 

NodeType mBoundaryNodes [NUM_DIMENSIONS][2]

Vector< Object > mPotentialScopeInObjects 

Object mReferenceObject 

NodeType * mTrackingList [NUM_DIMENSIONS][2]

ScopeTracker()

debugDump()

getReferenceObject()

registerObject(Object object)

setReferenceObject(Object object)

unregisterObject(Object object)

updateObject(Object object)