Resolution and representation of constraints on geometric and evolutive objects

The representation and propagation of functional and geometric const constraints in a CAD system seem to be two important components of future systems. Two approaches have been developed concurrently in our laboratory: The first one uses variational geometry. The geometric links between the objects in a CAD/CAM model will be represented in constraints graphs and then the properties of graphs, especially searches, will be used to modify and enrich the information dynamically. In order to modify the model cleverly, the study of the graph takes into account the notions of history and dynamic priorities class associated with the constraints. The implementation of this work is integrated in the software Sacado (Système Adaptatif de Conception Assistée et de Développement par Ordinateur), which has been realized at the LRIM. The second approach is based on object-oriented models using artificial intelligence tools. In the first realization, the constraints were represented immediately in Prolog; this point of view has been encouraged by the use of Smalltalk/V and Prolog/V. The present implementation tends to use C++ on a SUN workstation and a large part written in Smalltalk can be reused. A point of interest in developing these approaches concurrently is to compare them from different points of view (functional, performance, evolutivity, etc. without staying general criteria which are usually associated with object-oriented languages.