Skip to content
Sayiir

Tutorial: Order Processing (Python)

Build an order processing pipeline that validates orders, charges payment with retries, checks inventory in parallel, and finalizes the result — all with durable checkpointing.

from sayiir import Flow, run_durable_workflow, task


@task
def validate_order(order: dict) -> dict:
    if order["amount"] <= 0:
        raise ValueError("Invalid amount")
    return {**order, "validated": True}


@task(timeout="30s", retries=3)
def charge_payment(order: dict) -> dict:
    return {**order, "payment_id": "pay_123", "charged": True}


@task
def check_inventory(order: dict) -> dict:
    return {**order, "in_stock": True}


@task
def finalize(results: dict) -> str:
    payment = results["charge_payment"]
    inventory = results["check_inventory"]
    return f"Order complete: paid={payment['charged']}, stock={inventory['in_stock']}"


workflow = (
    Flow("order-processing")
    .then(validate_order)
    .fork()
        .branch(charge_payment)
        .branch(check_inventory)
    .join(finalize)
    .build()
)


status = run_durable_workflow(workflow, "order-1", {"order_id": 1, "amount": 99.99})

Demonstrates sequential tasks, fork/join parallelism, retries with timeouts, and durable execution. Swap InMemoryBackend for PostgresBackend to persist across restarts.

Full example: Order Processing (Python) Complete source code with instructions.

Next Steps

Section titled “Next Steps”