Doctest that perform I/O operations

Questions / Help
1

I’m building a very simple application, every time you call ModuleX.get(id) the system perform the folowing actions:

  1. Search for id on database
  2. If sucess, return the id’s related value
  3. Otherwise create a default data structure and save it on the database under the id key

The problem is my docTests are failing because I use the same id on all of them. Because of it, operations on former tests impact on the result of the later ones, because of the data was persisted

Questions:

  1. Am I using doctests wrong?
  2. There is a way to perform a delete operation after each doctest?
  3. How did you guys would approach this problem?
2
  1. this seems very ambitious for a doctest.
  2. are you using ecto sandboxes? I’m not 100% sure, but It should clean up after each doctest group. Try separating each of your cases into distinct ``` groups.
1 Like
3
  1. No, there’s nothing wrong with it.
  2. Yes, as mentioned by @ityonemo, Ecto’s sandbox adapter helps. What’s inside your test/test_helper.exs ?
1 Like
4

Actually my database is a very simple module that does binary parser and write it to disk using file name as identifier, I did not learn ecto yet.

I managed to make it work using the setup macro as following:

  setup do
    # Get the pids of all currently alive processes
    accounts_used_pids =
      DynamicSupervisor.which_children(Account.Cache)
      |> Stream.map(fn entry ->
        case entry do
          {_, pid, :worker, [Account.Server]} -> pid
          _ -> nil
        end
      end)
      |> Enum.filter(fn ele -> ele !== nil end)
    
    # Terminate all processes
    Enum.each(accounts_used_pids, &Process.exit(&1, :clean_up))
    
    # Reset the "database"
    File.rm_rf(Account.Database.folder_path())
    File.mkdir_p!(Account.Database.folder_path())

    :ok
  end

It might not be the optimal solution, but work for me! Thank you so much for the help!

5

Actually my database is a very simple module that does binary parser and write it to disk using file name as identifier, I did not learn ecto yet.

I managed to make it work using the setup macro as following:

  setup do
    # Get the pids of all currently alive processes
    accounts_used_pids =
      DynamicSupervisor.which_children(Account.Cache)
      |> Stream.map(fn entry ->
        case entry do
          {_, pid, :worker, [Account.Server]} -> pid
          _ -> nil
        end
      end)
      |> Enum.filter(fn ele -> ele !== nil end)
    
    # Terminate all processes
    Enum.each(accounts_used_pids, &Process.exit(&1, :clean_up))
    
    # Reset the "database"
    File.rm_rf(Account.Database.folder_path())
    File.mkdir_p!(Account.Database.folder_path())

    :ok
  end

It might not be the optimal solution, but work for me! Thank you so much for the help!

6

I managed to make it work using the setup macro as following:

  setup do
    # Get the pids of all currently alive processes
    accounts_used_pids =
      DynamicSupervisor.which_children(Account.Cache)
      |> Stream.map(fn entry ->
        case entry do
          {_, pid, :worker, [Account.Server]} -> pid
          _ -> nil
        end
      end)
      |> Enum.filter(fn ele -> ele !== nil end)
    
    # Terminate all processes
    Enum.each(accounts_used_pids, &Process.exit(&1, :clean_up))
    
    # Reset the "database"
    File.rm_rf(Account.Database.folder_path())
    File.mkdir_p!(Account.Database.folder_path())

    :ok
  end

The setup macro runs the code inside it after each test case. So I clean up the database and terminate all processes related to the account, after every test case.

It might not be the optimal solution, but work for me! Thank you so much for the help!