Summary
Workflow blocks orchestrate async steps; schedules run actions on an interval; commands define CLI entrypoints for point run.
workflow
Multi-step async orchestration. See examples/workflow.point and examples/workflow-retry.point for retry, timeout, and on failure policies.
schedule
Periodic jobs that call actions on an interval:
module Jobs
action health check
output status: Text
touches none
return "ok"
schedule health check tick
every 5 minutes
call health checkSee examples/tools/health-check-schedule.point.
For production cron, prefer the host scheduler and call a command or HTTP route — schedules emit setInterval for development and demos.
Job queue pattern
Use durable SQL rows for queue entries, `route` for enqueue/list/detail HTTP, `label`/pointJsonResponse when routes need branching status codes (such as conditional 404s), `workflow` to execute multi-step state transitions (pending → running → completed), and `command serve` plus `navigation` for a runnable demo shell. Illustration: `examples/app/job-queue/job-queue.point`. Interval ticks are a separate pattern; see `examples/tools/health-check-schedule.point`.
command
CLI entrypoints for point run:
point run prefers zero-argument command blocks, then main, then other zero-arg entrypoints. See examples/command.point and examples/app/todo.point.
command greet
output message: Text
return "Hello from Point"Common mistakes
- Defining a run entrypoint with required inputs (run needs zero-arg command/action/calculation)
- Forgetting
awaitbetween workflow steps that call actions
