module Marmot

Overview

Marmot is a non concurrent scheduler.

Marmot schedules tasks on three possibles ways:

• on a periodic span (Marmot.repeat)
• every day at a given time (Marmot.cron)
• when a value is available on a channel (Marmot.on)

Tasks are all executed on the same fiber. This means two things: first, you don't have to worry about concurrency (your tasks can share objects which does not support concurrency, like HTTP::Client), and second, they must not block (too much). If you want to execute jobs concurrently, you must spawn a new fiber inside your tasks.

A task receive a unique parameter which is an object representing itself. It can be canceled with Marmot::Task#cancel, from inside or outside the task. A canceled task can never be started again.

Tasks do not start when created. Instead, the main entrypoint is Marmot.run, which blocks while there are tasks to run. If there is no tasks to run, or they are all canceled, it stops.

The blocking behavior can also be stopped by calling Marmot.stop. As Marmot.run is blocking, you probably want to call Marmot.stop from a task or from another fiber.

When stopped, the tasks are not canceled and they will run again if Marmot.run is called again. To cancel all the tasks there is Marmot.cancel_all_tasks.

Extended Modules

• Marmot

Defined in:

Constant Summary

Log = ::Log.for(self)

VERSION = "0.2.2"

Instance Method Summary

• #cancel_all_tasks : Nil

  Cancels all the tasks.

• #cron(hour, minute, second = 0, &block : Callback) : Task

  Runs a task every day at hour and minute.

• #on(channel, &block : Callback) : Task

  Runs a task when a value is received on a channel.

• #repeat(span : Time::Span, first_run = false, &block : Callback) : Task

  Runs a task every given span.

• #run : Nil

  Starts scheduling the tasks.

• #stop

  Stops scheduling the tasks.

Instance Method Detail

channel = Channel(Int32).new
Marmot.on(channel) { |task| puts task.as(OnChannelTask).value }