This directory is a deterministic scenario benchmark for moon-fsm. It isdesigned to make the project scope and behavior reviewable without dependingon wall-clock timing, CPU model, or a network service.
The checked-in CSV contains 44 representative workflow cases across fourcommon engineering domains:
approval and review workflows
order fulfillment and cancellation
device boot, fault, and recovery flows
support ticket assignment, escalation, reopening, and closure
The extended support cases also cover customer waiting, cancellation,terminal-event storms, and long reopen loops. The runnable examples extendthose scenarios with examples/support_workflow for customer operations andexamples/incident_workflow for incident commander assignment, mitigation,rollback, journal ingestion, SLA checks, runbook handover, and closure review.
Each case records the initial state, initial context, event sequence, expectedfinal state, successful transition count, and rejected-event count. Theexecutable runner in main.mbt uses the same cases as typed data and verifiesthe expected history length and audit-entry count after every scenario.
Run the benchmark from the module root:
moon run benchmarks
This is a behavioral scenario benchmark, not a claim about throughput acrosshardware. It complements the unit tests by exercising longer, realistic pathsand explicit failure/recovery branches.