C# functional programming: how to treat IO in CRUD?

Question

For context, I recently discovered functional programming, and am trying to incorporate it into an existing Web API solution.

So far, I have been reading Functional Programming in C# and I've been trying to create a new controller that would be able to do basic CRUD using the language-ext library.

I'm running into a bit of an obstacle as I am not sure if I am keeping to FP principles when accessing the database.

Here is the code I have so far--please note that I am using Entity Framework Code First, and that for reasons beyond my control I am forced to use surrogate keys with type Guid, which is why there are so many operations involved in the method:

    // These functions are injected into the controller constructor
    // and are curried where appropriate.
    Func<int, Either<Error, Widget>> FetchWidgetById; // Impure
    Func<Widget, Widget, Either<Error, Widget>> CloneWidget;
    Func<Widget, Either<Error, Widget>> SaveToDb; // Impure
    Func<Either<Error, Widget>, IHttpActionResult> CreateHttpResponse

    public IHttpActionResult Update(Widget updatedWidget)
        => CreateHttpResponse(
            GetWidgetById(updatedWidget.HumanReadableId)
                .Bind(CloneWidget(updatedWidget))
                .Bind(SaveToDb));

The problem is that I am not sure how to signal that GetWidgets and UpdateDatabase are impure. I have come across the concept of the I/O monad but I don't think such a thing exists in the language-ext API, and I don't have enough FP knowledge to work out if there is an equivalent solution or name for it.

EDIT: I've found some documentation on monads on a similar but older project by the same author of language-ext. It's interesting to note that this project actually includes the I/O monad for which he acknowledges that

The IO monad may be seen as unnecessary in C# where everything has side-effects...

Answer 1

I'm not an expert in any of this, but I'll give it a shot in any case. My understanding is that if you want to have pure functions for domain logic and handle database access outside of this, you'll have to write quite a lot of boilerplate code. You'll have to have domain logic implemented as free monads, compose functionality using them, and finally have an interpreter to run the code, actually touching the database.

Since you are using language-ext, please refer to its documentation about free monads: https://github.com/louthy/language-ext/wiki/Thinking-Functionally:-Application-Architecture#free-monad

See also these examples:

• https://gist.github.com/louthy/524fbe8965d3a2aae1b576cdd8e971e4
• https://gist.github.com/dadhi/59cfc698f6dc6e31b722cd804aae185a
• https://gist.github.com/tonymorris/7817335

But, if you are just looking to signal to readers where code has side effects, it is not really possible in C#, because you could do IO anywhere, as was mentioned in your edit. Of course, being systematic and following conventions is possible, but I don't think you can go much beyond that. This applies to free monads as well, since nothing prevents from writing side effects there either in C#.