can somebody link me real world example of using a...

# arrow

[Marko Novakovic]

can somebody link me real world example of using arrow and functional programming? not something that just uses Either and calls itself functional programming nor something too complicated for purposes of showing theory in practice.

[simon.vergauwen]

Hey Marko,

What would you consider a real world example? For me that means “back-end”, and that typically is mostly suspend + Either with some concurrency/parallelism on top.

Sometimes Validated to accumulate errors when validating a form, or Ior when I need to figure out all failures and all success.

Besides that I use Arrow Fx Coroutines a lot, there is a lot of useful operators to compose suspend in parallel, or repeating/retrying with compose-able Schedules, compose-able Resources (concurrent/parallel), etc

[stojan]

maybe this helps: https://github.com/arrow-kt/frdomain.kt

it's based on an example from a Scala book ported to Kotlin and Arrow

[simon.vergauwen]

I’m also releasing a example back-end project soon. Using GCP, TestContainers, Ktor, Arrow (Core/Fx), Kotest, Exposed, etc

[thanh]

@simon.vergauwen cant wait for that 🙏

[simon.vergauwen]

I’m trying to clean it up a bit, and share the first version. It’s something I hope to keep improving as a working example of Arrow in the backend.

[Marko Novakovic]

@stojan thank you

[Marko Novakovic]

@simon.vergauwen I should’ve made myself more clear. I want some app, Android or backed that is not just CRUD app, has some more logic to it not just write/read from database or network requests. some library that’s written FP way etc.

[Marko Novakovic]

looking forward to example you mentioned

[Ivan Dugalic]

Hi @Marko Novakovic I’m sharing this library that is focusing on functional domain modeling (optimized for event-sourcing and CQRS). It could inspire you 😉 https://github.com/fraktalio/fmodel It was inspired by Arrow and it is using Arrow on the application layer.

[Ivan Dugalic]

Sharing some demos/usage https://github.com/fraktalio/fmodel-demos It is all very new, but I have learned a lot in the process of adopting functional style. Kotlin is wonderful, and Kotlin idiomatic Arrow is a cherry on the top 😉

[carbaj0]

I will dedicate August to study this 👨‍🎓

[Marko Novakovic]

@Ivan Dugalic thank you very much!

[raulraja]

@Ivan Dugalic Fraktalio is awesome!, thanks for sharing this. Does Flow ever show up to model anything in there? Many apps that use functional domain modeling in places like Scala and others end up relying in fs2 for streams or have some kind of data type to represent their streams and interactions with other systems. I was wondering how would you connect something like a Kafka stream with Fraktalio following this arch.

[Ivan Dugalic]

Thank you @raulraja Domain lib (<https://github.com/fraktalio/fmodel/tree/main/domain>) is pure computation and behaviour. The Application lib (<https://github.com/fraktalio/fmodel/tree/main/application>) is more opinionated in regarding to the runtime target. At the moment, I am proposing couple of interfaces (eg. <https://github.com/fraktalio/fmodel/blob/main/application/src/main/kotlin/com/fraktalio/fmodel/application/EventRepository.kt>) with `suspend`ing only. I guess that Flow API is something that can be added as well. Thanks for suggestion!

[Ivan Dugalic]

I am also hoping that this will inspire people to create they own Application and Infrastructure layers (with or without Reactive API)

[simon.vergauwen]

I’ve been very interested in doing something similar, but based on what I’m used to from functional streaming.

A Flow<A> or a sequence of suspend () -> A composes in all situations, and suspend () -> A only composes for single element effects.
So Flow<A> can also seen as a “abstraction” over suspend () -> A where you don’t know the amount of elements it will result in,
when using such approaches I’ve seen that it almost always results in nicer/better leaner apps.

Since you gain control over the sequencing of suspension, you can gain more power over pull-based things.
It becomes much easier to add rate limiters etc, which is composition around a sequence of suspend () -> A.

[simon.vergauwen]

It’s all also achievable with suspend () -> A, but I found that in places where you need to deal with things such as rate limiters the resulting code always gets rather complex while it can be implemented as a generic operation over Flow<A>.

[Ivan Dugalic]

Hi @simon.vergauwen Yes, indeed! You are right. Adding Flow/Flux to the `suspend () -> A`/`Mono` is an important step to more robust streaming API.

[simon.vergauwen]

Looking forward to see this evolve!

[Ivan Dugalic]

Thanks again for your feedback. I appreciate it!

[Ivan Dugalic]

Thinking/Learning Kotlin Flow exception handling (in relation to Arrow Either).

[Ivan Dugalic]

interface EventRepository<C, E> {
    suspend fun C.fetchEvents1(): Either<Error.FetchingEventsFailed, Flow<E>>
    fun C.fetchEvents2(): Flow<Either<Error.FetchingEventsFailed, E>>
    fun C.fetchEvents3(): Flow<E>
}

[Ivan Dugalic]

What approach to take 🙂 ?

[Ivan Dugalic]

Option 3 sounds good. Option 2 looks promising.

[Ivan Dugalic]

Option 1 is naive 🙂 I would say

[simon.vergauwen]

Well these 3 are completely different 😄

1. I think this signature is super rare. It’s when building the reference of Flow<E> requires side-effects, and it can fail with Error.FetchingEventsFailed
2. This is probably the most promising signature yes. It’s

[simon.vergauwen]

Grrrr Slack constantly breaks formatting across ` and I get into an incorrect state

[simon.vergauwen]

The 3rd one you have no explicit domain errors, but that could be valid depends on the use-cases.
2 and 3 will play nice together anyway 🙂

[simon.vergauwen]

The first one you can probably drop the suspend also by wrapping that suspend into Flow before you actually emit things

[simon.vergauwen]

flow { setUp() }.concatFlatMap { flowOfEvents() }

[Ivan Dugalic]

Thanks Simon. I will stick with 2 and 3….and going with the Flow. Iterable is no more present.

[simon.vergauwen]

Feel free to share the PR here, I’d love to make some time to take a look

[Ivan Dugalic]

Essentially, it will be one Flow from fetching current Events/State from the repo till the moment you save new Events/State.

[Ivan Dugalic]

I am implementing it now, and it will be ready till the end of the week most probably…(being busy with other stuff )

[Ivan Dugalic]

Will share of course, thanks again!

[simon.vergauwen]

Sounds really nice. Always happy to help 🙂

[Ivan Dugalic]

Switching from Iterable to Flow PR <https://github.com/fraktalio/fmodel/pull/33>. It is still work in progress and sorry for one big commit 😞 (it make sense to browse the branch directly: <https://github.com/fraktalio/fmodel/tree/feature/flow>)

[Ivan Dugalic]

@simon.vergauwen I was in hurry moving this forward. The change is major, and it will upgrade the lib version to 2.0.0. I plan to merge it soon and experiment with the API publicly (increasing minor versions)

[Jörg Winter]

@Ivan Dugalic
👏 for the plans for 2.0
Is the 2.0.0-SNAPSHOT stable enough, to give upgrading the fmodel-demos a try ?

PS: thanks for using Maven 😄

[Ivan Dugalic]

It is stable, please use it and share the experience. Upgrading the fmodel-demos would be awesome!!! Thanks!