Refined types

[Cotel]

Close #1099

The goal of this PR is to provide a Refinement<F, A> typeclass for creating refined types and also offering a basic set of them.

[Cotel]

If I add a method fun nonZero(a: A): Kind<F, A> = a.nonZero() the IDE tells me its signature clashes with this method 🤔

[Cotel]

I'm trying to create the same refining mechanism as in https://github.com/fthomas/refined but I don't think I'm doing it right or that if that will work for Kotlin 🤔

I've defined interface GreaterEqual<F, A> : Not<F, A, Less<F, A>> and it needs an overriden Refinement which Not will use to negate the refinement. The problem is that, as you can see in GreaterEqualTest you need to pass a Less instance when doing x.greaterEqual() and I there is no place where I can get that instance.

Besides, GreaterEqual in the Scala's refined library is just a typealias and I was expecting to do the same but it seems that is not possible. Maybe it is even better to define this refined type from 0 like Less instead of extending Not.

[Cotel]

I would like to have a function with the same signature as less(Ord<Int>, max: Int) instead of passing a Less instance 🤷‍♂️

[pakoito]

We wanted to do away with arrow-validation soon. Do you mind keeping this open for a few more days while we decide?

[Cotel]

Sure there is no hurry 👌

[raulraja]

Arrow validation has to change from the current use of Either to be a validation library similar in scope to refined. I think this code can be merged in as is as we evolve and deprecate other functions in Arrow Validation which for now can just be ignored for this development.

[Cotel]

I've been trying to implement Modulo and Positive but I'm having some troubles.

For Modulo I need to have access to the Int.rem(x: Int): Int operation but as I'm working with Modulo<F, A: Number> Number has not access to this operation. I would need A to be a concrete type but I would have to write lots of instances (ValidatedModuloInt, EitherModuloInt, ValidatedModuloFloat, EitherModuloFloat...)

For Positive I can define it like interface Positive<F, A : Number> : Greater<F, A> and then override the fun min(): A function from Greater to always return 0. But it depends on the concrete type again which 0 to return. In Scala's refined library they make use of _0 from shapeless but I cannot do something similar in Kotlin or at least I don't know how to do it.

For the moment I think I will stick to the refined types defined in https://github.com/gcanti/newtype-ts which I think I can port them all even though they are very few and silly. Besides, I don't like the 101.greaterEqual(Less(100)) which I told in my previous comments so I will remove the Not refined type and remake GreaterEqual and LessEqual to be more like their non-Equal types. Once I have them implemented we can work from there.

Tell me your opinions pls.

[raulraja]

Sounds like a good path forward, let me know when ready for review.

[Cotel]

All the refined types (besides their Integer specific version) from fp-ts are implemented now. This is ready for review.

Which should be our next step? Should I write some documentation on the usage and the definition of custom refined types or should I try to implement more types from the scala library? I would like to stay in this branch if there is no hurry in finishing this 😄

[raulraja]

@Cotel This looks great, outstanding work 👏 Let's add the kdocs with runnable examples in all public api's and this is good to merge. See the AQL or Bracket Kdoc for an example.

[raulraja]

Should we make these operator?

[Cotel]

Do you mean infix functions? I can overload the not operator which I suppose it would be ! but I cannot overload greater as an operator 🤔

[raulraja]

should we make these operator?

[raulraja]

All of these are lawful type classes. We should consider including the laws of each one of them or just for Refinement for now.
That will give users a framework to test their new refined rules that are composed with multiple refinements. No need to be in this PR but would be great to have.

[Cotel]

I'm trying to run dokka but it tells me that import arrow.validation.xxx <- unresolved reference: validation for every refined type 🤔

[Cotel]

I still don't know whats going on with dokka/ank. I have tried to compile the arrow-validation project in the arrow-docs project but then ank won't parse a single file. Could anyone look into this? The branch is almost finished except for the docs.

[1Jajen1]

Ank fails because the import is not correct. It should be import arrow.validation.refinedTypes.numeric.validated.nonZero.nonZero (or import arrow.validation.refinedTypes.numeric.validated.nonZero.*) instead. A good way to test snippets like this is to create a small class (top level or something to not implicitly import anything) and then actually try to compile and run the code in the snippets.

[Cotel]

Soo I've been trying to fix this for some days now but I have no idea of what to do. The only way I can fix the imports is by including arrow-validation as a dependency for arrow-docs but then Ank won't run on a single file. It fails on the first document it needs to generate. Any ideas ?

[raulraja]

@Cotel Can we see the output and command you are using?

[Cotel]

I'm doing the same the CI is doing (clean build dokka :arrow-docs:runAnk)

And its failing with this

> Task :arrow-docs:runAnk
�[35m      :::     ::::    ::: :::    :::
    :+: :+:   :+:+:   :+: :+:   :+:
   +:+   +:+  :+:+:+  +:+ +:+  +:+
  +#+     ++: +#+ +:+ +#+ +#++:++
  +#+     +#+ +#+  +#+#+# +#+  +#+
  #+#     #+# #+#   #+#+# #+#   #+#
  ###     ### ###    #### ###    ###�[0m
�[31m[50%] ✗ /home/travis/build/arrow-kt/arrow/modules/docs/arrow-docs/build/site/docs/effects/effect/README.md [1 of 2]�[0m
Exception in thread "main" 

"""
import arrow.reflect.*
import arrow.effects.typeclasses.*

TypeClass(Effect::class).dtMarkdownList()
"""

�[31mjava.lang.ExceptionInInitializerError
	at arrow.reflect.TypeClassKt.extensions(TypeClass.kt:17)
	at arrow.reflect.TypeClassKt.supportedDataTypes(TypeClass.kt:41)
	at arrow.reflect.ExtensionsHelperKt.dtMarkdownList(ExtensionsHelper.kt:31)
Caused by: java.lang.ClassNotFoundException: arrow.core.Eitherrrow.data.NonEmptyList
	at java.net.URLClassLoader.findClass(URLClassLoader.java:381)
	at java.lang.ClassLoader.loadClass(ClassLoader.java:424)
	at sun.misc.Launcher$AppClassLoader.loadClass(Launcher.java:335)
	at java.lang.ClassLoader.loadClass(ClassLoader.java:357)
	at java.lang.Class.forName0(Native Method)
	at java.lang.Class.forName(Class.java:264)
	at arrow.reflect.TypeClassExtensionKt.unKind(TypeClassExtension.kt:57)
	at arrow.reflect.TypeClassExtensionKt.asClassTypeExtension(TypeClassExtension.kt:25)
	at arrow.reflect.TypeClassExtensionKt.<clinit>(TypeClassExtension.kt:51)
	at arrow.reflect.TypeClassKt.extensions(TypeClass.kt:17)
	at arrow.reflect.TypeClassKt.supportedDataTypes(TypeClass.kt:41)
	at arrow.reflect.ExtensionsHelperKt.dtMarkdownList(ExtensionsHelper.kt:31)
	at Line_1.<init>(Unknown Source)
	at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
	at sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:62)
	at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
	at java.lang.reflect.Constructor.newInstance(Constructor.java:423)
	at org.jetbrains.kotlin.cli.common.repl.GenericReplEvaluator.eval(GenericReplEvaluator.kt:95)
	at org.jetbrains.kotlin.cli.common.repl.GenericReplCompilingEvaluator.eval(GenericReplCompilingEvaluator.kt:78)
	at org.jetbrains.kotlin.cli.common.repl.GenericReplCompilingEvaluator.compileAndEval(GenericReplCompilingEvaluator.kt:53)
	at org.jetbrains.kotlin.cli.common.repl.ReplAtomicEvalAction$DefaultImpls.compileAndEval$default(ReplApi.kt:168)
	at org.jetbrains.kotlin.cli.common.repl.KotlinJsr223JvmScriptEngineBase.compileAndEval(KotlinJsr223JvmScriptEngineBase.kt:61)
	at org.jetbrains.kotlin.cli.common.repl.KotlinJsr223JvmScriptEngineBase.eval(KotlinJsr223JvmScriptEngineBase.kt:31)
	at javax.script.AbstractScriptEngine.eval(AbstractScriptEngine.java:264)
	at arrow.ank.InterpreterKt$monadDeferInterpreter$1$compileCode$$inlined$run$lambda$1$1.invokeSuspend(interpreter.kt:142)
	at arrow.ank.InterpreterKt$monadDeferInterpreter$1$compileCode$$inlined$run$lambda$1$1.invoke(interpreter.kt)
	at arrow.typeclasses.Monad$binding$wrapReturn$1.invokeSuspend(Monad.kt:56)
	at kotlin.coroutines.jvm.internal.BaseContinuationImpl.resumeWith(ContinuationImpl.kt:32)
	at kotlin.coroutines.ContinuationKt.startCoroutine(Continuation.kt:127)
	at arrow.typeclasses.Monad$DefaultImpls.binding(Monad.kt:57)
	at arrow.typeclasses.MonadError$DefaultImpls.binding(MonadError.kt)
	at arrow.typeclasses.MonadThrow$DefaultImpls.binding(MonadError.kt)
	at arrow.effects.typeclasses.MonadDefer$DefaultImpls.binding(MonadDefer.kt)
	at arrow.effects.instances.IOMonadDeferInstance$DefaultImpls.binding(io.kt)
	at arrow.effects.instances.io.monadDefer.IOMonadDeferInstanceKt$monadDefer$1.binding(IOMonadDeferInstance.kt:83)
	at arrow.typeclasses.MonadContinuation.binding(MonadContinuations.kt)
	at arrow.ank.InterpreterKt$monadDeferInterpreter$1$compileCode$$inlined$run$lambda$1.invokeSuspend(interpreter.kt:133)
	at kotlin.coroutines.jvm.internal.BaseContinuationImpl.resumeWith(ContinuationImpl.kt:32)
	at arrow.typeclasses.MonadContinuation$bind$$inlined$suspendCoroutineUninterceptedOrReturn$lambda$1.invoke(MonadContinuations.kt:61)
	at arrow.typeclasses.MonadContinuation$bind$$inlined$suspendCoroutineUninterceptedOrReturn$lambda$1.invoke(MonadContinuations.kt:15)
	at arrow.effects.IO$flatMap$1.invoke(IO.kt:89)
	at arrow.effects.IO$flatMap$1.invoke(IO.kt:34)
	at arrow.effects.IORunLoop.step(IORunLoop.kt:120)
	at arrow.effects.IO.unsafeRunTimed(IO.kt:124)
	at arrow.effects.IO.unsafeRunSync(IO.kt:121)
	at arrow.ank.main.main(main.kt:14)�[0m

> Task :arrow-docs:runAnk FAILED

[raulraja]

I believe @nomisRev fixed a very similar issue on his branch recently. @nomisRev do you know if your changes made it to master? @Cotel is seeing the same issue you had with dtMarkDownList

[raulraja]

@Cotel let's get this merged in once CI passes and continue with more granular features. More action and contributions are coming to arrow-validated and we should make sure all this great work is considered. Thoughts?

[Cotel]

I think this is a good start. We can merge and continue from this point 👍