Self-paced loop

A polling or work loop whose next iteration and delay depend on observed job state rather than a fixed calendar schedule.

A monitor can query a durable status endpoint, wait with backoff, and stop on a terminal success, failure, cancellation, or deadline. The application should interpret explicit states rather than ask a model to infer completion from vague text. Event callbacks may be more efficient when the service supports them.

Builder example

Adaptive polling suits variable-duration jobs but still needs rate limits, maximum duration, cancellation, transient-error handling, and a durable state for the final result. The loop should not hold a user interaction open indefinitely or hide an exhausted failure.

Common confusion: A cron schedule fires at a fixed clock time whether or not anything changed. A self-paced loop reacts to the state of one job and decides its own next check based on what it sees, which suits work whose length varies from run to run.