Abstract

Every computer program yields a number of questions. The big one is "Does it work?", but more specific tasks of debugging and optimization pose a wide variety of separate questions. In an ideal world, automatic tools would provide the programmers with these answer, but unfortunately the famous theorem by Rice tells us that this is only possible for completely trivial questions. Rather than giving up, programming language researchers have deconstructed the exact premises of Rice's theorem to change the ground rules sufficiently to allow useful results. In fact, any weakening of these premises that is logically possible has yielded an active area of research. This talk will survey these areas and zoom in on some extreme points in the design space of solutions.