TaskAfter

Your Libraries & OS Mentoring Libraries
1

This was requested a number of times in the past few days, so I whipped it up.

This fully supports both a global config mode as well as per-application registration so all forms should be supported.

I make no claims as to reliability or bugs or anything of the sort, so please report any that you found. However it is fully passing its test suite currently, which is this:

  test "TaskAfter and forget" do
    s = self()
    {:ok, _auto_id} = TaskAfter.task_after(500, fn -> send(s, 42) end)
    assert_receive(42, 600)
  end

  test "TaskAfter and receive" do
    {:ok, _auto_id} = TaskAfter.task_after(500, fn -> 42 end, send_result: self())
    assert_receive(42, 600)
  end

  test "TaskAfter with custom id" do
    assert {:ok, :my_id} = TaskAfter.task_after(500, fn -> 42 end, id: :my_id, send_result: self())
    assert_receive(42, 600)
  end

  test "TaskAfter with custom id duplicate fails" do
    assert {:ok, :dup_id} = TaskAfter.task_after(500, fn -> 42 end, id: :dup_id, send_result: self())
    assert {:error, {:duplicate_id, :dup_id}} = TaskAfter.task_after(500, fn -> 42 end, id: :dup_id, send_result: self())
    assert_receive(42, 600)
  end

  test "TaskAfter lots of tasks" do
    assert {:ok, _} = TaskAfter.task_after(400, fn -> 400 end, send_result: self())
    assert {:ok, _} = TaskAfter.task_after(200, fn -> 200 end, send_result: self())
    assert {:ok, _} = TaskAfter.task_after(500, fn -> 500 end, send_result: self())
    assert {:ok, _} = TaskAfter.task_after(100, fn -> 100 end, send_result: self())
    assert {:ok, _} = TaskAfter.task_after(300, fn -> 300 end, send_result: self())
    assert {:ok, _} = TaskAfter.task_after(600, fn -> 600 end, send_result: self())
    assert_receive(100, 150)
    assert_receive(200, 150)
    assert_receive(300, 150)
    assert_receive(400, 150)
    assert_receive(500, 150)
    assert_receive(600, 150)
  end

  test "TaskAfter non-global by name" do
    assert {:ok, pid} = TaskAfter.Worker.start_link(name: :testing_name)
    {:ok, _auto_id} = TaskAfter.task_after(500, fn -> 42 end, send_result: self(), name: :testing_name)
    assert_receive(42, 600)
    GenServer.stop(pid)
  end

  test "TaskAfter non-global by pid" do
    assert {:ok, pid} = TaskAfter.Worker.start_link()
    assert {:ok, _auto_id} = TaskAfter.task_after(500, fn -> 42 end, send_result: self(), pid: pid)
    assert_receive(42, 600)
    GenServer.stop(pid)
  end

  test "TaskAfter in process (unsafe, can freeze the task worker if the task does not return fast)" do
    assert {:ok, pid} = TaskAfter.Worker.start_link()
    s = self()
    assert {:ok, _auto_id} = TaskAfter.task_after(500, fn -> send(s, self()) end, send_result: :in_process, pid: pid)
    assert_receive(^pid, 600)
    GenServer.stop(pid)
  end
19 Likes
2

New version, added a TaskAfter.cancel_task_after/2 as well and hardened up some tests for if callbacks are poorly made and crash out (even in_process is safe now as long as they do not take ‘long’).
The new test suite:

defmodule TaskAfterTest do
  use ExUnit.Case, async: true
  doctest TaskAfter

  test "TaskAfter and forget" do
    s = self()
    assert {:ok, _auto_id} = TaskAfter.task_after(500, fn -> send(s, 42) end)
    assert_receive(42, 600)
  end

  test "TaskAfter and receive" do
    assert {:ok, _auto_id} = TaskAfter.task_after(500, fn -> 42 end, send_result: self())
    assert_receive(42, 600)
  end

  test "TaskAfter with custom id" do
    assert {:ok, :my_id} = TaskAfter.task_after(500, fn -> 42 end, id: :my_id, send_result: self())
    assert_receive(42, 600)
  end

  test "TaskAfter with custom id duplicate fails" do
    assert {:ok, :dup_id} = TaskAfter.task_after(500, fn -> 42 end, id: :dup_id, send_result: self())
    assert {:error, {:duplicate_id, :dup_id}} = TaskAfter.task_after(500, fn -> 42 end, id: :dup_id, send_result: self())
    assert_receive(42, 600)
  end

  test "TaskAfter lots of tasks" do
    assert {:ok, _} = TaskAfter.task_after(400, fn -> 400 end, send_result: self())
    assert {:ok, _} = TaskAfter.task_after(200, fn -> 200 end, send_result: self())
    assert {:ok, _} = TaskAfter.task_after(500, fn -> 500 end, send_result: self())
    assert {:ok, _} = TaskAfter.task_after(100, fn -> 100 end, send_result: self())
    assert {:ok, _} = TaskAfter.task_after(300, fn -> 300 end, send_result: self())
    assert {:ok, _} = TaskAfter.task_after(600, fn -> 600 end, send_result: self())
    assert_receive(100, 150)
    assert_receive(200, 150)
    assert_receive(300, 150)
    assert_receive(400, 150)
    assert_receive(500, 150)
    assert_receive(600, 150)
  end

  test "TaskAfter non-global by name" do
    assert {:ok, pid} = TaskAfter.Worker.start_link(name: :testing_name)
    {:ok, _auto_id} = TaskAfter.task_after(500, fn -> 42 end, send_result: self(), name: :testing_name)
    assert_receive(42, 600)
    GenServer.stop(pid)
  end

  test "TaskAfter non-global by pid" do
    assert {:ok, pid} = TaskAfter.Worker.start_link()
    assert {:ok, _auto_id} = TaskAfter.task_after(500, fn -> 42 end, send_result: self(), pid: pid)
    assert_receive(42, 600)
    GenServer.stop(pid)
  end

  test "TaskAfter in process (unsafe, can freeze the task worker if the task does not return fast)" do
    assert {:ok, pid} = TaskAfter.Worker.start_link()
    s = self()
    assert {:ok, _auto_id} = TaskAfter.task_after(500, fn -> send(s, self()) end, send_result: :in_process, pid: pid)
    assert_receive(^pid, 600)
    GenServer.stop(pid)
  end

  test "TaskAfter and cancel timer, do not run the callback" do
    cb = fn -> 42 end
    assert {:ok, auto_id} = TaskAfter.task_after(500, cb)
    assert {:ok, ^cb} = TaskAfter.cancel_task_after(auto_id)
  end

  test "TaskAfter and cancel timer, but its already been run or does not exist" do
    assert {:error, {:does_not_exist, :none}} = TaskAfter.cancel_task_after(:none)
    assert {:ok, auto_id} = TaskAfter.task_after(0, fn -> 42 end, send_result: self())
    assert_receive(42, 100)
    assert {:error, {:does_not_exist, ^auto_id}} = TaskAfter.cancel_task_after(auto_id)
  end

  test "TaskAfter and cancel but also run the callback in process (unsafe again)" do
    assert {:ok, auto_id} = TaskAfter.task_after(500, fn -> 42 end)
    assert {:ok, 42} = TaskAfter.cancel_task_after(auto_id, run_result: :in_process)
  end

  test "TaskAfter and cancel but also run the callback async" do
    s = self()
    assert {:ok, auto_id} = TaskAfter.task_after(500, fn -> send(s, 42) end)
    assert {:ok, :task} = TaskAfter.cancel_task_after(auto_id, run_result: :async)
    assert_receive(42, 600)
  end

  test "TaskAfter and cancel but also run the callback async while returning result to pid" do
    s = self()
    assert {:ok, auto_id} = TaskAfter.task_after(500, fn -> 42 end)
    assert {:ok, :task} = TaskAfter.cancel_task_after(auto_id, run_result: s)
    assert_receive(42, 600)
  end

  test "TaskAfter and crash" do
    s = self()
    len = &length/1
    d = len.([])
    assert {:ok, _auto_id2} = TaskAfter.task_after(100, fn -> send(s, 21) end)
    assert {:ok, _auto_id1} = TaskAfter.task_after(250, fn -> send(s, 1/d) end)
    assert {:ok, _auto_id2} = TaskAfter.task_after(500, fn -> send(s, 42) end)
    assert_receive(42, 600)
    assert_receive(21, 1)
    assert :no_message == (receive do m -> m after 1 -> :no_message end)
  end
end
3 Likes