Revision Recognition for Scientific Computing: Theory and Application.
Andrew Strelzoff · 2003
Scientific computing applications are rarely equipped with high quality graphic user interfaces (GUIs), in part because the interfaces cannot be revised quickly enough to keep pace with rapid changes in the underlying application. A critical step in user interface maintenance is revision recognition: the process of determining, given two versions of a program, which parts of the original program have been changed, deleted or preserved. For large, highly complex scientific applications this problem is a significant barrier to interface maintenance. In this paper we present the "reticent programmer model" which transforms the revision recognition problem into an optimization problem. The solutions obtained using this model are exact for small applications and are high quality estimates for medium to large applications. An example of how these solutions can be used to facilitate user interface maintenance is also presented.