How do functional languages handle a mocking situation when using Interface based design?

Question

Typically in C# I use dependency injection to help with mocking;

public void UserService
{
 public UserService(IUserQuery userQuery, IUserCommunicator userCommunicator, IUserValidator userValidator)
 {
 UserQuery = userQuery;
 UserValidator = userValidator;
 UserCommunicator = userCommunicator;
 }
 ...
 public UserResponseModel UpdateAUserName(int userId, string userName)
 {
 var result = UserValidator.ValidateUserName(userName)
 if(result.Success)
 {
 var user = UserQuery.GetUserById(userId);
 if(user == null)
 {
 throw new ArgumentException();
 user.UserName = userName;
 UserCommunicator.UpdateUser(user);
 }
 }
 ...
 }
 ...
}
public class WhenGettingAUser
{
 public void AndTheUserDoesNotExistThrowAnException()
 {
 var userQuery = Substitute.For<IUserQuery>();
 userQuery.GetUserById(Arg.Any<int>).Returns(null);
 var userService = new UserService(userQuery);
 AssertionExtensions.ShouldThrow<ArgumentException>(() => userService.GetUserById(-121));
 }
}

Now in something like F#: if I don't go down the hybrid path, how would I test workflow situations like above that normally would touch the persistence layer without using Interfaces/Mocks?

I realize that every step above would be tested on its own and would be kept as atomic as possible. Problem is that at some point they all have to be called in line, and I'll want to make sure everything is called correctly.

• Robert Harvey: In theory, unit testing in a functional language is a much easier proposition than OOP testing, because functions merely return values (i.e. no side effects) and there's no mutable state.
• Robert Harvey: @ElYusubov: Please don't put tags in question titles; see meta.stackexchange.com/questions/128548/… for more information.
• Robert Harvey: I repeat: don't put tags in question titles. We already have a tagging system, and it already has the F# tag in it.
• Yusubov: Just saw your comment. Will not put tags for further edits, Thanks !

Answer 1

Typically, I would go about this by writing a function that returns a representation of the action that I want to perform on the external resource rather than performing the operation itself. This way the logic and the act of IO are decoupled and testing is much easier.

Answer 2

The same way you do in C#, but instead of passing the mock into the constructor, you pass it to the function itself (or as a part of a parameter passed to the function).