Next: Monitoring the Progress of
Up: growing-point
Previous: Repair Mechanisms
When a well-formed GPL program is running an Amorphous computation,
one of two results can usually be expected. Either the computation
will terminate with the accomplishment of the goal, or a growing-point
will become "stuck" midway through a computation. A growing point may
become stuck for a variety of reasons, but ultimately, it is because
none of the neighboring values received by the growing point meets the
criterion for propagation. Because the growing point cannot progress
towards its immediate goal, we have a simple example of the failure of
a module to achieve conditions promised by the plan. The question
that naturally arises is how we may extend the Growing Point Language
to recognize when it has become ``stuck,'' probe its neighbors to
determine what may have caused the failure, and implement a new
strategy based on the information it found.
Jeremy Zucker
2000-06-10