A Visualization Tool for Constraint Program Debugging

Abstract

Constraint programming is an emerging technology that offers an original approach allowing for efficient and flexible solving of complex problems. Its main advantage relies in its ability to compute with partial information expressed in terms of constraints. These constraints are monotonically, accumulated during the program execution in order to restrict the problem search space. We address one of the cornerstones of this technology, namely the current lack of debugging facilities. In particular visualization and understanding of the underlying constraint system during program executions is very important. We propose to structure this huge, flat and intricate part of the execution data in order to provide access to high level examination of its evolution. More precisely, we present a means to hierarchically organize sets of constraints in order to divide them into manageable parts while presenting computation correctness. Soundness of our method is shown, an algorithm supporting it is given, and an implemented prototype exhibiting its effectiveness is described.