Generic computation of bulletin boards into geometric kernels

Nowadays, many commercial CAD systems are built on proprietary geometric kernel which provide an API containing a set of high level geometric operations (boolean operations, slot, chamfering, etc). Because of their complexity, these operations can generate important modifications on topological cells (vertices, edges, faces, volumes, etc.) of the objects. At the same time, many of these kernels need to know precisely what has occurred to each topological cell belonging to objects given or resulting from a previous high level geometric operation. At the end of each operation, the geometric kernel must provide a bulletin board describing cells' evolution through a list of events (split, merge, creation, deletion). Most commercial geometric kernels use B-Rep structures and provide methods enabling the developer of a CAD system to retrieve a number of events that occurred on cells. These kernels have their own scheme for detecting events, based on their own taxonomy of situations, heuristics and evolution rules. Little is known of their details, which are proprietary information, let alone of the underlying theory, if any. Generally, for example, the detected events are not generic for all cells' dimensions. This lack of underlying theory limits the possibility to extend the use of these kernels to new domains of investigation. In this paper, we propose a generic model that enables to create a bulletin board. This bulletin board will contain the complete list of events having occurred on cells of any dimension, and that belong to any topological model. The genericity of this model and the completeness in all dimensions of this list are based on the use of four elementary mechanisms (split_elem, merge_elem, crea_elem, del_elem). They are defined independently of the topological model, and allow the generation of the bulletin board, whatever the geometric operation. This model has been implemented using the geometric kernel of the modeler Moka, based on generalized maps.