We consider pragmatic issues in applying constraint-based theories (such as that developed for data exchange) to a variety of problems. We identify disconnects between theoreticians and tool developers, and propose principles for creating problem units that are appropriate for tools. Our Downstream Principle then explains why automated schema mapping is a prerequisite to transitioning schemamatching prototypes. Next, we compare concerns, strengths, and weaknesses of database and AI approaches to data exchange. Finally, we discuss how constrained update by business processes is a central difficulty in maintaining n-tier applications, and compare the challenges with data exchange and conventional view update