Kotlin type inference on "supposedly" right types - kotlin

I am new to Kotlin and I was playing with it. I pretty much wanted to create a pretty basic event bus. So I came up with this
interface Event
interface EventListener<E : Event> {
fun handle(event: E)
}
interface EventBus {
fun <E : Event> registerListener(aClass: Class<E>, eventListener: EventListener<E>)
}
class MyBus() : EventBus {
private val eventListeners: MutableMap<String, MutableList<EventListener<out Event>>> = mutableMapOf()
constructor(listeners: List<Pair<Class<Event>, EventListener<Event>>>) : this() {
listeners.forEach {
registerListener(it.first, it.second)
}
}
override fun <E : Event> registerListener(aClass: Class<E>, eventListener: EventListener<E>) {
val key = aClass.name
val listeners: MutableList<EventListener<out Event>> = eventListeners.getOrPut(key) { mutableListOf() }
listeners.add(eventListener)
}
}
val bus = MyBus(
listOf(
MyEvent::class.java to MyEventListener()
)
)
class MyEvent : Event
class AnotherEvent : Event
class MyEventListener : EventListener<MyEvent> {
override fun handle(event: MyEvent) {
}
}
what happens is that when I try to create MyBus using the constructor accepting the list of pairs, I get
Type inference failed. Expected type mismatch: inferred type is List<Pair<Class<MyEvent>,MyEventListener>> but List<Pair<Class<Event>,EventListener<Event>>> was expected
But if I change the constructor to be something like
constructor(listeners: List<Pair<Class<out Event>, EventListener<out Event>>>) : this() {
listeners.forEach {
registerListener(it.first, it.second)
}
}
adding out pretty much everywhere, then the MyBus constructor works, but the invocation to registerListener(..) breaks for the same exact reason as before. So the only way to solve this is to add "out"s also on registerListener function.
I suspect I'm doing something wrong here, but I don't know what precisely. Any help?

If you want your EventListener to be able to consume Events, then its type has to be invariant or covariant (not declared out). If it let you pass your EventListener<MyEvent> as if it were an EventListener<Event>, then your MyBus class might call listener.handle(event) on it with some Event that is not a MyEvent, such as AnotherEvent. Then you will get a ClassCastException when it tries to cast this AnotherEvent to MyEvent.
To be able to store different types of invariant EventHandlers, you will have to remove the variance restrictions by using star projection, and cast them when you retrieve them from the map. So make the map keys into class objects instead of just Strings. Since you will not have the help of the compiler when working with the star-projected types, you need to be careful that you are only adding an item to your MutableMap that is of the same type as the Class key that's associated with it. Then when you retrieve items, only cast to an invariant type.
The other part of your issue is that your constructor needs a generic type. Right now it works exclusively with Event so it can't handle subtypes of Event. Kotlin doesn't (yet?) support generic types for constructors so you have to do this with a factory function.
Here's an example of all the above.
class MyBus() : EventBus {
private val eventListeners: MutableMap<Class<*>, MutableList<EventListener<*>>> = mutableMapOf()
override fun <E : Event> registerListener(aClass: Class<E>, eventListener: EventListener<E>) {
val listeners = retrieveListeners(aClass)
listeners.add(eventListener)
}
private fun <E: Event> retrieveListeners(aClass: Class<E>): MutableList<EventListener<E>> {
#Suppress("UNCHECKED_CAST")
return eventListeners.getOrPut(aClass) { mutableListOf() } as MutableList<EventListener<E>>
}
}
// Factory function
fun <E : Event> myBusOf(listeners: List<Pair<Class<E>, EventListener<E>>>): MyBus {
return MyBus().apply {
listeners.forEach {
registerListener(it.first, it.second)
}
}
}
And you might want to change the type of the factory parameter from a <List>Pair to a vararg Pair so it's easier to use.
Here's a stripped down example to explain the variance limitation.
Your interface for an Event consumer:
interface EventListener<E : Event> {
fun handle(event: E)
}
Two implementations of Event:
class HelloEvent: Event {
fun sayHello() = println("Hello world")
}
class BoringEvent: Event {}
A class implementing the interface:
class HelloEventListener: EventListener<HelloEvent> {
override fun handle(event: HelloEvent) {
event.sayHello()
}
}
Now you have an EventListener that can handle only HelloEvents. Try to treat it like an EventListener<Event>:
val eventListener: EventListener<Event> = HelloEventListener() // COMPILE ERROR!
Imagine the compiler did not prevent you from doing this and you do this:
val eventListener: EventListener<Event> = HelloEventListener()
eventListener.handle(BoringEvent()) // CLASS CAST EXCEPTION AT RUN TIME!
If this were allowed your HelloEventListener would try to call sayHello() on the BoringEvent, which doesn't have that function, so it will crash. This is what generics are here to protect you from.
Now suppose your HelloEventListener.handle() didn't call event.sayHello(). Well, then it could have safely handled a BoringEvent. But the compiler isn't doing that level of analysis for you. It just knows what you declared, that HelloEventListener cannot handle anything except HelloEvent.

Related

Kotlin context receivers cannot resolve members of generic type

In the following code, the call member of Animal cannot be resolved even though Cat is specified as context receiver and it has a member named call.
interface Animal { val call: String }
object Cat : Animal { override val call: String = "Meow" }
object Dog : Animal { override val call: String = "Woof" }
fun <T : Animal> acquireAnimal(animal: T, block: context(T) () -> Unit) {
block(animal)
}
fun main() {
acquireAnimal(Cat) {
call
}
}
When I type this inside the lambda, then the IDE seems to suggest that the type of this is Any?.
If I do the same with a function without a generic context receiver, then it seems to get the type right.
Is this a limitation that is by design or is this a bug?
The fact that you cannot access call was a bug, which was fixed in Kotlin 1.7.20.
A workaround for lower versions is:
sealed interface TypeWrapper<out A> {
object IMPL: TypeWrapper<Nothing>
}
fun <T: Animal> acquireAnimal(animal: T, block: context(T) (TypeWrapper<T>) -> Unit) {
block(animal, TypeWrapper.IMPL)
}
fun main() {
acquireAnimal(Cat) {
val x = call // works!
}
}
However, the fact that this doesn't work is intended. Context receivers do not change the meaning of this. Since you are in a global function, this does not mean anything, and the existence of a context receiver does not change that.
Normally, to access the context receiver itself, you need to do a qualified this by appending the generated label for the context receiver:
context(Foo)
fun foo() {
val x = this#Foo
}
However, your context receiver is a type parameter, so according to the rules here, I don't think a label is generated for the context receiver.

Can I omit type in generics? - Kotlin

If I have a following interface:
interface BaseDataRemote<T, in Params> {
fun getData(params: Params? = null): Single<T>
}
Would it be possible have implementation of this interface that does not take Params?
To have effectively something like:
interface BaseDataRemote<T> {
fun getData(): Single<T>
}
Implementation is as follows:
class RemoteSellerDataSource #Inject constructor(
private val sellerApi: SellerApi,
#Named("LANG") private val lang: String
) : BaseDataRemote<SellerEntity, Nothing> {
override fun getData(params: Nothing?): Single<SellerEntity> {
return sellerApi.getSeller(lang).map { it.fromApiEntity() }
}
}
I use Dagger 2 to module to bind this implementation:
#Module
internal interface RemoteModule {
#Binds
#CoreScope
fun bindsSellerRemote(remoteSellerDataSource: RemoteSellerDataSource): BaseDataRemote<SellerEntity, Nothing>
}
I tried using Nothing as second type parameter, but it does not seem to work
(I'm getting required: class or interface without bounds error
Full error message:
RemoteSellerDataSource.java:6: error: unexpected type
public final class RemoteSellerDataSource implements com.bigchangedev.stamps.business.sdk.data.base.data.BaseDataRemote<SellerEntity, ?> {
^
required: class or interface without bounds
found:?
Thanks.
EDIT: the original answer was a pure Kotlin answer because the OP didn't mention Dagger.
Using Nothing is correct and works in pure Kotlin. However, Dagger seems to convert your code to Java, and in doing so it uses wildcards for the generics (which it doesn't like because it wants exact type matches). To avoid this issue, you can try using #JvmSuppressWildcards on your generic type parameters:
class RemoteSellerDataSource #Inject constructor(
private val sellerApi: SellerApi,
#Named("LANG") private val lang: String
) : BaseDataRemote<SellerEntity, #JvmSuppressWildcards Nothing> {
override fun getData(params: Nothing?): Single<SellerEntity> {
return sellerApi.getSeller(lang).map { it.fromApiEntity() }
}
}
Although I'm not sure what will happen in Java with Nothing in that case. I guess this should have the same effect on the Java code as removing the in variance for the second type param in the interface declaration, but without weakening your Kotlin types.
Another workaround would be to use Unit instead of Nothing, which Dagger will most likely convert to Void in this case. This is not great for your types, though.
Original answer:
You can technically already call getData() without arguments thanks to the default value. An implementation that doesn't care about the params argument can simply expect null all the time.
The Kotlin type that only contains null and no other value is technically Nothing?, and since getData is defined with Params? (note the ?) as input, it should be correct to specify Nothing (even without ?) as second type argument. So you should be able to define an implementation like this:
interface BaseDataRemote<T, in Params> {
fun getData(params: Params? = null): Single<T>
}
class ImplementationWithoutParams<T> : BaseDataRemote<T, Nothing> {
override fun getData(params: Nothing?): Single<T> {
// params will always be null here
}
}
To avoid confusion for the users, this implementation may additionally provide a getData() method without arguments at all:
class ImplementationWithoutParams<T> : BaseDataRemote<T, Nothing> {
override fun getData(params: Nothing?): Single<T> = getData()
fun getData(): Single<T> {
TODO("implementation")
}
}

Kotlin generics supertype not applied

I was coding on Java for quite a long time and trying to migrate to Kotlin. I'm confused with Generics in Kotlin a bit...
I have a DelegateManager class. It should consume only subtypes of IViewData
class DelegateManager<T : IViewData> {
private val delegates: MutableList<AdapterDelegate<T>> = mutableListOf()
fun addDelegate(adapterDelegate: AdapterDelegate<T>) {
delegates.add(adapterDelegate)
}
...
}
Inside TrackListAdapter I want to add a delegate. As you might have seen it's AdapterDelegate<TrackViewData> and TrackViewData is a subtype of IViewData So it should work but it shows error inside init block of TrackListAdapter
class TrackListAdapter : BaseListAdapter<IViewData>() {
init {
delegateManager.addDelegate(TrackViewDelegate()) // error: Type mismatch -> Required: AdapterDelegate<IViewData>, Found: TrackViewDelegate
}
}
class TrackViewDelegate : AdapterDelegate<TrackViewData>() {
override fun onCreateViewHolder(parent: ViewGroup): ListViewHolder<TrackViewData> {
val itemView = LayoutInflater.from(parent.context).inflate(R.layout.track_item, parent, false)
return TrackViewHolder(itemView)
}
override fun isDelegateForDataType(data: IViewData) = data is TrackViewData
}
How to deal with it? How to extend the generic parameter correctly?

multiple instances of superclass member attribute

I have an abstract stub class that needs to be subclasses to add spring annotations.
abstract class ConsumerStub<TYPE>{
val receivedMessages: MutableMap<String, TYPE> = ConcurrentHashMap()
open fun processMessage(#Payload payload: TYPE, record: ConsumerRecord<String, *>) {
this.receivedMessages[record.key()] = payload
}
fun receivedMessageWithKey(key: String): Boolean = this.receivedMessages.contains(key)
fun receivedMessageWithKeyCallable(key: String): Callable<Boolean> = Callable { receivedMessageWithKey(key) }
fun getReceiveMessageWithKey(key: String): TYPE? = this.receivedMessages[key]
fun reset() {
this.receivedMessages.clear()
}
}
for example:
open class WorkflowRequestConsumerStub: ConsumerStub<InternalWorkflowRequest>() {
#KafkaListener(
id = "xyzRequestConsumerStub",
topics = ["abc-workflow-requests"]
)
override fun processMessage(
#Payload payload: InternalWorkflowRequest,
record: ConsumerRecord<String, *>
) {
super.processMessage(payload, record)
}
}
I am seeing some really weird behaviour with the receivedMessages.
After some debugging I realised there seem to be 2 instances of receivedMessages.
stub.reset() throws a null pointer exception
after changing the code to initialize receivedMessages in reset(), processMessage() and receivedMessageWithKey() are seeing 2 different receivedMessages with different objectIds.
what's going on? In java the subclass should have access to any protected members in the super, so I assume the same applies to kotlin.
UPDATE: this works as expected when defining receivedMessages abstract and override in the implementations. That really sucks if this is how it should be done in kotlin. In this case there is no need for the implementation to care about the map.

How to specify "own type" as return type in Kotlin

Is there a way to specify the return type of a function to be the type of the called object?
e.g.
trait Foo {
fun bar(): <??> /* what to put here? */ {
return this
}
}
class FooClassA : Foo {
fun a() {}
}
class FooClassB : Foo {
fun b() {}
}
// this is the desired effect:
val a = FooClassA().bar() // should be of type FooClassA
a.a() // so this would work
val b = FooClassB().bar() // should be of type FooClassB
b.b() // so this would work
In effect, this would be roughly equivalent to instancetype in Objective-C or Self in Swift.
There's no language feature supporting this, but you can always use recursive generics (which is the pattern many libraries use):
// Define a recursive generic parameter Me
trait Foo<Me: Foo<Me>> {
fun bar(): Me {
// Here we have to cast, because the compiler does not know that Me is the same as this class
return this as Me
}
}
// In subclasses, pass itself to the superclass as an argument:
class FooClassA : Foo<FooClassA> {
fun a() {}
}
class FooClassB : Foo<FooClassB> {
fun b() {}
}
You can return something's own type with extension functions.
interface ExampleInterface
// Everything that implements ExampleInterface will have this method.
fun <T : ExampleInterface> T.doSomething(): T {
return this
}
class ClassA : ExampleInterface {
fun classASpecificMethod() {}
}
class ClassB : ExampleInterface {
fun classBSpecificMethod() {}
}
fun example() {
// doSomething() returns ClassA!
ClassA().doSomething().classASpecificMethod()
// doSomething() returns ClassB!
ClassB().doSomething().classBSpecificMethod()
}
You can use an extension method to achieve the "returns same type" effect. Here's a quick example that shows a base type with multiple type parameters and an extension method that takes a function which operates on an instance of said type:
public abstract class BuilderBase<A, B> {}
public fun <B : BuilderBase<*, *>> B.doIt(): B {
// Do something
return this
}
public class MyBuilder : BuilderBase<Int,String>() {}
public fun demo() {
val b : MyBuilder = MyBuilder().doIt()
}
Since extension methods are resolved statically (at least as of M12), you may need to have the extension delegate the actual implementation to its this should you need type-specific behaviors.
Recursive Type Bound
The pattern you have shown in the question is known as recursive type bound in the JVM world. A recursive type is one that includes a function that uses that type itself as a type for its parameter or its return value. In your example, you are using the same type for the return value by saying return this.
Example
Let's understand this with a simple and real example. We'll replace trait from your example with interface because trait is now deprecated in Kotlin. In this example, the interface VitaminSource returns different implementations of the sources of different vitamins.
In the following interface, you can see that its type parameter has itself as an upper bound. This is why it's known as recursive type bound:
VitaminSource.kt
interface VitaminSource<T: VitaminSource<T>> {
fun getSource(): T {
#Suppress("UNCHECKED_CAST")
return this as T
}
}
We suppress the UNCHECKED_CAST warning because the compiler can't possibly know whether we passed the same class name as a type argument.
Then we extend the interface with concrete implementations:
Carrot.kt
class Carrot : VitaminSource<Carrot> {
fun getVitaminA() = println("Vitamin A")
}
Banana.kt
class Banana : VitaminSource<Banana> {
fun getVitaminB() = println("Vitamin B")
}
While extending the classes, you must make sure to pass the same class to the interface otherwise you'll get ClassCastException at runtime:
class Banana : VitaminSource<Banana> // OK
class Banana : VitaminSource<Carrot> // No compiler error but exception at runtime
Test.kt
fun main() {
val carrot = Carrot().getSource()
carrot.getVitaminA()
val banana = Banana().getSource()
banana.getVitaminB()
}
That's it! Hope that helps.
Depending on the exact use case, scope functions can be a good alternative. For the builder pattern apply seems to be most useful because the context object is this and the result of the scope function is this as well.
Consider this example for a builder of List with a specialized builder subclass:
open class ListBuilder<E> {
// Return type does not matter, could also use Unit and not return anything
// But might be good to avoid that to not force users to use scope functions
fun add(element: E): ListBuilder<E> {
...
return this
}
fun buildList(): List<E> {
...
}
}
class EnhancedListBuilder<E>: ListBuilder<E>() {
fun addTwice(element: E): EnhancedListBuilder<E> {
addNTimes(element, 2)
return this
}
fun addNTimes(element: E, times: Int): EnhancedListBuilder<E> {
repeat(times) {
add(element)
}
return this
}
}
// Usage of builder:
val list = EnhancedListBuilder<String>().apply {
add("a") // Note: This would return only ListBuilder
addTwice("b")
addNTimes("c", 3)
}.buildList()
However, this only works if all methods have this as result. If one of the methods actually creates a new instance, then that instance would be discarded.
This is based on this answer to a similar question.
You can do it also via extension functions.
class Foo
fun <T: Foo>T.someFun(): T {
return this
}
Foo().someFun().someFun()