Why use Kind instead of F

Question

Whenever you read Arrow.kt tutorials, as they progress, they start touting as a benefit of FP that you can abstract away your types and write everything as things like fun <F> myFun(): Kind<F, A> instead of fun myFun(): IO<A> or fun myFun(): Option<A>.

I understand what Kind<F, A> means, but I don't understand why doing this is so beneficial. It seems to me what is being enabled is that your function could return an Option or an Either or an IO without having to change your types, but is that really such a big benefit?

I've never used FP in team projects or anything featuring hundreds of thousands of lines of code, so maybe refactoring this can be troublesome and Kind<F, A> means you don't have to change functional signatures, just the code in the blocks, I'm not sure.

Can someone give me some concrete examples of this being useful in practice? An example where doing this has yielded some big benefit?

Like I can understand having an interface like Logger and have a file logger, API-pinging logger, STDOUT logger, etc. implement and you can swap out, but is this the same thing? How?

They cannot introduce an F generic variable that takes another generic variable, so they had to introduce `Kind` and `.fix()` to overcome this. — EpicPandaForce, Jan 07 '20 at 20:29
Why is having that F generic variable useful, is what I'm asking. I understand the technical aspect of why Kind is needed over F, but why not just use Option or Either or whatever? — user1713450, Jan 07 '20 at 20:32
I can understand internally why it'd be useful because they can more easily write a single function for all Functors or something if you're an Arrow developer, but why is it touted in advanced Arrow examples as a thing you want to progress toward? Tutorials will often tout that you can interchange your types after you do this, but when would you as an end-user really need to swap out Option for IO or Deferred ? They all serve vastly different purposes. — user1713450, Jan 07 '20 at 20:33
@user1713450 Being able to swap them out is one advantage. Being able to write generic functions yourself (not only when you're an Arrow developer, but also when just composing some arrows) is another. But I think the most important one is that you can use the same familiar functions on all these types, and will know what to write as soon as your realise the concrete type you're operating on is a functor. — Bergi, Jan 07 '20 at 20:41
Why `Option` in the first place when we just can use `Option`, `Option`? Because polymorphism allows generalized expressions and higher kinded polymorphism allows even more generalized ones and that is a property people usually aspire to. — , Jan 07 '20 at 21:01
Check out [this article](https://sidburn.github.io/blog/2016/03/24/higher-kinded-polymorphism) (uses F# and C# for the examples) but the same principles apply. TL;DR it's all about code reuse — LordRaydenMK, Jan 07 '20 at 21:09
@lordraydenmk you just electricuted my brain. Fatality! :) That makes sense. Looks like I just haven't worked on anything FP complex enough to benefit from this if it's about code reuse. — user1713450, Jan 07 '20 at 23:16
@Bergi if you make your comment an answer instead, I'll mark it the correct one. There were good, helpful comments from others, but yours was self-contained and clear. Now that I've done another month of FP and advanced, I can tell your answer is a good one since I've been trying to code in final tagless occasionally. — user1713450, Feb 16 '20 at 21:02

Why use Kind instead of F

0 Answers0