Turbine

Appearance

Getting Started ​

Install ​

bash
go get github.com/YakirOren/turbine

Quick Start ​

go
package main

import (
	"context"
	"log"

	"github.com/YakirOren/turbine"
)

func main() {
	rt := turbine.NewStandalone(turbine.Config{})
	defer rt.Shutdown()

	processOrder := func(ctx turbine.Context, orderID string) (string, error) {
		// Each step's result is recorded in SQLite.
		// On crash, the workflow resumes from the last completed step.
		chargeID, err := turbine.Do(ctx, func(ctx context.Context) (string, error) {
			return "ch_" + orderID, nil
		}, turbine.WithStepName("charge"))
		if err != nil {
			return "", err
		}

		return turbine.Do(ctx, func(ctx context.Context) (string, error) {
			return "shipped " + orderID + " (charge: " + chargeID + ")", nil
		}, turbine.WithStepName("ship"))
	}
	turbine.Register(rt, processOrder)

	if err := rt.Launch(); err != nil {
		log.Fatal(err)
	}

	handle, _ := turbine.Run(rt, processOrder, "ord-42")
	result, _ := handle.GetResult()
	log.Println(result)
}

Serving HTTP Alongside Workflows ​

If you want to trigger workflows from HTTP routes, use turbine.NewApp instead. It returns both the app (for routing, Start, etc.) and the runtime:

go
app, rt := turbine.NewApp(turbine.Config{})

turbine.Register(rt, processOrder)

app.OnServe().BindFunc(func(e *turbine.ServeEvent) error {
	e.Router.POST("/orders/{id}", func(re *turbine.RequestEvent) error {
		handle, _ := turbine.Run(rt, processOrder, re.Request.PathValue("id"))
		result, _ := handle.GetResult()
		return re.JSON(200, map[string]string{"result": result})
	})
	return e.Next()
})

log.Fatal(app.Start())

What Happens on First Launch ​

What happens on first launch

On first launch, Turbine automatically:

  1. Creates the SQLite collections it needs (pt_workflow_status, pt_operation_outputs, etc.)
  2. Recovers any pending workflows from a previous run
  3. Starts the queue runner and cron scheduler

What's Next? ​