I have a behaviours with lots of functions. Each pass an arguments and an optional keyword list of options. And I found myself in this weird situation where testing with Mox introduce some unsatisying results.

A reduction of the problem :

defmodule MyBehaviour do
  @callback greet(name :: String.t(), greeting :: String.t()) :: String.t()
end

defmodule MyModule do
  @behaviour MyBehaviour

  @impl true
  def greet(name, greeting \\ "Hello") do
    "#{greeting}, #{name}!"
  end
end

And a test like this

import Mox

defmock(MyMock, for: MyBehaviour)

MyMock
|> expect(:greet, fn name -> "Hello #{name}" end)

result = MyMock.greet("World")

It fails because the mock does not know any function of arity 1 called greet.

** (ArgumentError) unknown function greet/1 for mock MyMock
    (mox 1.2.0) lib/mox.ex:681: Mox.add_expectation!/4
    (mox 1.2.0) lib/mox.ex:549: Mox.expect/4

If I test instead

import Mox

defmock(MyMock, for: MyBehaviour)

MyMock
|> expect(:greet, fn name, _ -> "Hello #{name}" end)

result = MyMock.greet("World")

It obviously fails, because the function with one argument is never mocked.

** (UndefinedFunctionError) function MyMock.greet/1 is undefined or private. Did you mean:

      * greet/2

    MyMock.greet("World")

Which leaves me with two choices (I think) :

1. Duplicating all callback with and without the optional arguments (sad when there is a lot of callbacks)
2. Removing the “optional” from the arguments and always calling the two arity functions.

I chose option 2 with a lack of enthusiasm, anyone in the same situation chose a different solution ? Or had a better way of dealing with the situation ?