使用 state[:pid] 传递的消息失败，新生成的 pid 有效

Question

我的第二次测试失败了。

 1) test it jumps when it can (RunnerTest)
     test/runner_test.exs:15
     No message matching %{y: 1} after 100ms.
     The process mailbox is empty.
     stacktrace:
       test/runner_test.exs:18: (test)

defmodule RunnerTest do
  use ExUnit.Case
  doctest Runner

  setup_all do
    {:ok, pid: spawn(fn -> Runner.input() end)}
  end

  test "it increases its x position", state do
    avatar = %{x: 0, y: 0}
    send state[:pid], {:run, self, avatar}
    assert_receive %{x: 1}
  end

  test "it jumps when it can", state do
    avatar = %{x: 0, y: 0}
    send state[:pid], {:jump, self, avatar}
    assert_receive %{y: 1}
  end

end

跑步者模块：

defmodule Runner do
    def input do
      receive do
       {:run, sender, mover} -> send sender, run(mover)
       {:jump, sender, mover} -> send sender, jump(mover)
     end
    end

    defp run(mover) do
        Map.merge(mover, %{x: mover.x + 1})
    end

    defp jump(mover) do
        case {mover.y} do
          {0} ->
            Map.merge(mover, %{y: mover.y + 1})
          _ ->
            mover
        end
    end

end

此测试通过：

 test "it jumps when it can", state do
    avatar = %{x: 0, y: 0}
    pid = spawn(fn -> Runner.input() end)
    send pid, {:jump, self, avatar}
    assert_receive %{y: 1}
  end

为什么 1 个测试 (运行) 使用设置而不是另一个（跳转）？

Answer 1

Why is 1 test (run) working with setup but not the other (jump)?

这是因为您的进程只响应一条消息然后就死掉了，而您使用的是 setup_all 而不是 setup，这意味着该进程只启动一次，在第一次测试之前运行，并且相同的 pid 传递给所有测试。

如果您想为每个测试生成一个新进程，您可以将 setup_all 更改为 setup，或者让您的进程响应多于 1 条消息（使用自递归可能无限）。

使用 state[:pid] 传递的消息失败，新生成的 pid 有效

Message passing with state[:pid] fails, freshly generated pid works

elixir

ex-unit