In the specification of value classes, it says:
A value class can only extend universal traits and cannot be extended itself. A universal trait is a trait that extends Any, only has defs as members, and does no initialization. Universal traits allow basic inheritance of methods for value classes, but they incur the overhead of allocation. For example
trait Printable extends Any {
def print(): Unit = println(this)
}
class Wrapper(val underlying: Int) extends AnyVal with Printable
val w = new Wrapper(3)
w.print() // actually requires instantiating a Wrapper instance
First Question
Now, I would take this to mean that the following (probably) does not require instantiation:
trait Marker extends Any
class Wrapper(val underlying: Int) extends AnyVal with Marker {
def print(): Unit = println(this) //unrelated to Marker
}
val w = new Wrapper(3)
w.print() //probably no instantiation as print is unrelated to Marker
Am I correct?
Second Question
And I would think there is an even chance as to whether this requires instantiation or not:
trait Printable extends Any {
def print(): Unit //no implementation
}
class Wrapper(val underlying: Int) extends AnyVal with Printable {
override def print() = println(this) //moved impl to value class
}
val w = new Wrapper(3)
w.print() // possibly requires instantiation
On the balance of probability, I would also think that no instantiation would be needed - am I correct?
Edit
I'd not thought about the exact implementation of print() in the example:
def print(): Unit = println(this)
Let's say that I used the following instead:
def print(): Unit = println(underlying)
Would these cause instantiations?
Am I correct?
No, we can see it if we emit the final compilation output with -Xprint:jvm:
<synthetic> object F$Wrapper extends Object {
final def print$extension($this: Int): Unit =
scala.Predef.println(new com.testing.F$Wrapper($this));
This is due to the fact println has a type signature requiring Any, so we're shooting ourselves in the foot here since we're effectively "treating the value class ttpe as another type".
Although the call is dispatched to the static method call:
val w: Int = 3;
F$Wrapper.print$extension(w)
We're still incurring the allocation inside print$extension.
If we stray away from using Wrapper.this, then your first assumption is indeed correct and we can see the compiler happily unwrap Wrapper:
<synthetic> object F$Wrapper extends Object {
final def print$extension($this: Int): Unit =
scala.Predef.println(scala.Int.box($this));
And the call site now looks like this:
val w: Int = 3;
com.testing.F$Wrapper.print$extension(w)
This is valid for both of your examples now, as there is no need for any dynamic dispatch on the created interface.
Related
I'm writing a type-safe code and want to replace apply() generated for case classes with my own implementation. Here it is:
import shapeless._
sealed trait Data
case object Remote extends Data
case object Local extends Data
case class SomeClass(){
type T <: Data
}
object SomeClass {
type Aux[TT] = SomeClass { type T = TT }
def apply[TT <: Data](implicit ev: TT =:!= Data): SomeClass.Aux[TT] = new SomeClass() {type T = TT}
}
val t: SomeClass = SomeClass() // <------------------ still compiles, bad
val tt: SomeClass.Aux[Remote.type] = SomeClass.apply[Remote.type] //compiles, good
val ttt: SomeClass.Aux[Data] = SomeClass.apply[Data] //does not compile, good
I want to prohibit val t: SomeClass = SomeClass() from compiling. Is it possible to do somehow except do not SomeClass to be case class?
There is a solution that is usually used if you want to provide some smart constructor and the default one would break your invariants. To make sure that only you can create the instance you should:
prevent using apply
prevent using new
prevent using .copy
prevent extending class where a child could call the constructor
This is achieved by this interesing patten:
sealed abstract case class MyCaseClass private (value: String)
object MyCaseClass {
def apply(value: String) = {
// checking invariants and stuff
new MyCaseClass(value) {}
}
}
Here:
abstract prevents generation of .copy and apply
sealed prevents extending this class (final wouldn't allow abstract)
private constructor prevents using new
While it doesn't look pretty it's pretty much bullet proof.
As #LuisMiguelMejíaSuárez pointed out this is not necessary in your exact case, but in general that could be used to deal with edge cases of case class with a smart constructor.
UPDATE:
In Scala 3 you only need to do
case class MyCaseClass private (value: String)
and it will prevent usage of: apply, new and copy from outside of this class and its companion.
This behavior was ported to Scala 2.13 with option -Xsource:3 enabled. You have to use at least 2.13.2 as in 2.13.1 this flag doesn't fix the issue.
So you can make the constructor private and ensure that T is also something different to Nothing.
I believe the best way to ensure the constructor is private (as well as many other things as #MateuszKubuszok show) is to use a (sealed) trait instead of a class:
(if you can not use a trait for whatever reasons, please refer to Mateusz's answer)
import shapeless._
sealed trait Data
final case object Remote extends Data
final case object Local extends Data
sealed trait SomeClass {
type T <: Data
}
object SomeClass {
type Aux[TT] = SomeClass { type T = TT }
def apply[TT <: Data](implicit ev1: TT =:!= Data, ev2: TT =:!= Nothing): Aux[TT] =
new SomeClass { override final type T = TT }
}
Which works like this:
SomeClass() // Does not compile.
SomeClass.apply[Remote.type] // Compiles.
SomeClass.apply[Data] // Does not compile.
You can see it running here.
If you want to prohibit using some of auto-generated methods of a case class you can define the methods (with proper signature) manually (then they will not be generated) and make them private (or private[this]).
Try
object SomeClass {
type Aux[TT] = SomeClass { type T = TT }
def apply[TT <: Data](implicit ev: TT =:!= Data): SomeClass.Aux[TT] = new SomeClass() {type T = TT}
private def apply(): SomeClass = ??? // added
}
val t: SomeClass = SomeClass() // doesn't compile
val tt: SomeClass.Aux[Remote.type] = SomeClass.apply[Remote.type] //compiles
val ttt: SomeClass.Aux[Data] = SomeClass.apply[Data] //doesn't compile
In principle, the methods (apply, unapply, copy, hashCode, toString) can be generated not by compiler itself but with macro annotations. Then you can choose any subset of them and modify their generation as you want.
Generate apply methods creating a class
how to efficiently/cleanly override a copy method
Also the methods can be generated using Shapeless case classes a la carte. Then you can switch on/off the methods as desired too.
https://github.com/milessabin/shapeless/blob/master/examples/src/main/scala/shapeless/examples/alacarte.scala
https://github.com/milessabin/shapeless/blob/master/core/src/test/scala/shapeless/alacarte.scala
An example is worth a thousand words:
class A { def foo: Any = new Object }
class B extends A {
override def foo: AnyVal = 42
}
In Java, the signature #Override public int foo() wouldn't even be allowed, and the overridden method foo in B could only return the wrapper integer type (#Override java.lang.Integer foo()).
Is Scala able to avoid the boxing/unboxing of AnyVal values in the overridden def foo: AnyVal method above?
No, it does not. Scala has to adhere to emitting the correct bytecode:
λ scalac -Xprint:jvm Bar.scala
[[syntax trees at end of jvm]] // Bar.scala
package yuval.tests {
class A extends Object {
def foo(): Object = new Object();
def <init>(): yuval.tests.A = {
A.super.<init>();
()
}
};
class B extends yuval.tests.A {
override def foo(): Object = scala.Int.box(42);
def <init>(): yuval.tests.B = {
B.super.<init>();
()
}
}
}
You can see that although AnyVal was permitted in Scala, the actual method signature for the emitted foo is Object and not AnyVal, and Int is boxed.
Yuval's answer can be generalized: erasure of AnyVal is Object (you can see this e.g. by entering classOf[AnyVal] in the REPL), so whenever you have AnyVal in Scala, you can expect Object in bytecode.
E.g. if you change A to
class A { def foo: AnyVal = 0 }
it's still Object.
Maybe there is some situation in which using AnyVal itself will avoid boxing, but I would be surprised. It was created pretty much for compiler's convenience, and picked up another use (value classes) later, but it's rarely useful in user code (except for defining value classes).
How can I achieve this:
final case class ChairId(id: String)
trait GeneratorLike[TO, TC <: AbstractId] {
val prefix: String
def generate(): TC = TO.apply(prefix + "-" + UUID.randomUUID())
}
implicit object ChairIdGenerator extends GeneratorLike[ChairId.type, ChairId] {
val prefix: String = "CHAIR"
}
implicit def IdFn[TO, TC <: AbstractId](x: TO)(implicit ev: GeneratorLike[TO, TC]): GeneratorLike[TO, TC] = ev
//right now I can call:
ChairId.generate()
I don't want to define companion object for that situation and I wondered if there is a chance to extend object with use of implicits?
When I do (I use TO as TypeObject and TC as TypeClass naming) idFn[TO, TC] I want TO to be object that implements def apply(id: String): TC can I enforce that? And how would I get to use this function? It feels totally impossible to call function on type parameter :/
It is impossible to call a method on a type parameter, because it represents a type and not an object. You can call a method on an object, because it is something that exists, but a type is an abstract concept. I don't know what your motivation is for wanting to implicitly add generate() to companion objects, because it actually requires just as much code to define an implicit GeneratorLike than it does to define the companion for ChairId.
If you force GeneratorLike to have an apply method (which can be implemented by case class apply), and remove the first type parameter, this will work.
trait GeneratorLike[TC <: AbstractId] { this: Singleton =>
val prefix: String
def apply(id: String): TC
def generate(): TC = apply(prefix + "-" + UUID.randomUUID())
}
abstract class AbstractId
final case class ChairId(id: String) extends AbstractId
object ChairId extends GeneratorLike[ChairId] {
val prefix = "CHAIR"
}
scala> ChairId.generate()
res0: ChairId = ChairId(CHAIR-60bb01c7-af95-46c7-af45-0b3fa78b3080)
Structural typing is not a particularly good idea on the JVM, so always try to avoid the def test(x: {def apply(s: String)}): TC type stuff because it is implemented using reflection which can be a dog performance wise.
Second, you should probably avoid using val inside a trait. Read here.
The approach you have considered is actually the right one, and namely type classes.
trait HasGenerator[T] {
def apply(uuid: String): T
def generate[T : Generator] = apply(Generator[T].generate)
}
final case class ChairId(id: String)
object ChairId extends HasGenerator[ChairId]
trait Generator[TO] {
def prefix: String
def generate(): String = prefix + "-" + UUID.randomUUID()
def apply(): String = generate
}
object Generator {
def apply[T : Generator] = implicitly[Generator[T]]
}
// Notice .type is not necessary
implicit object ChairIdGenerator extends Generator[ChairId] {
override def prefix = "CHAIR"
}
Why not just use:
ChairId(Generator[ChairId])
This all seems like overkill though so you can quite easily somehow. It's worth fleshing out your requirements a bit more because type classes don't really seem super necessary just yet. You could just do with:
Update
If you use something like the HasGenerator that I have added above in conjunction with the companion object, you can now successfully call ChairId.generate()
I'd like to "lock" a class, which is extended from a trait. Is it possible in Scala?
For example I have:
trait A {
val boris: String
val john: String
val number: Int
}
class B extends A {
// do something with these values
}
but can I ensure, that in class B no new values will be added if those aren't declared in trait A?
Thanks for your answers.
You cannot.
But if you simply mark the trait as sealed and provide a default implementation:
sealed trait A { val boris: String }
final class B(val boris: String) extends A {}
then people are free to create implicit value classes that make it look like new functionality has been added (except without actually creating the class):
implicit class MyB(val underlying: B) extends AnyVal {
def sirob = underlying.boris.reverse
}
(new B("fish")).sirob // "hsif"
You can also let the classes take a type parameter as a marker if you want to keep them straight at compile-time (though not runtime):
sealed trait A[T] { val boris: String }
final class B[T](val boris: String) extends A[T] {}
implicit class MyB(val underlying: B[Int]) extends AnyVal {
def sirob = underlying.boris.reverse
}
(new B[Int]("fish")).sirob // "hsif"
(new B[Char]("fish")).sirob // error: value sirob is not a member of B[Char]
So you could--especially with 2.10--simply lock everything and let users enrich the original interface this way.
I'm not sure if this covers your intended use case, though; it doesn't provide any inheritance.
Based on your example and my guess at what you are actually trying to do, you may want to consider just using case classes.
Extending a case class is generally avoided (I think it will spit out deprecation warnings if you try), so that will prevent people from wanting to extend your class in order to add functionality.
Translating your example into a case class:
case class A (boris: String, john: String, number: Int)
Then instead of extending A to change its values, you'd just make a new instance, e.g.
val a2 = someOtherA.copy(john="Doe")
I have been working on an issue with implicit conversion for days now, but somehow I just cannot figure out what I am doing wrong. I read through all the other questions on SO that deal with implicits but I still don't understand what the problem is.
As an example, let's consider a Java interface like this(T extends Object for brevity):
public interface JPersistable<T extends Object> {
public T persist(T entity);
}
In scala, I do the following:
case class A()
case class B() extends A
case class C()
case class D() extends C
trait Persistable[DTOType <: A, EntityType <: C] {
// this would be implemented somewhere else
private def doPersist(source: EntityType): EntityType = source
// this does not implement the method from the Java interface
private def realPersist(source: DTOType)(implicit view: DTOType => EntityType): EntityType = doPersist(source)
// this DOES implement the method from the Java interface, however it throws:
// error: No implicit view available from DTOType => EntityType.
def persist(source: DTOType): EntityType = realPersist(source)
}
case class Persister() extends Persistable[B, D] with JPersistable[B]
object Mappings {
implicit def BToD(source: B): D = D()
}
object Test {
def main(args: Array[String]) {
import Mappings._
val persisted = Persister().persist(B())
}
}
As stated in the comment, I get an exception at compile time. I guess my questions are:
1) Why do I need to specify the implicit conversion on the doRealPersist explicitly? I expected the conversion to happen even if I do the following:
trait Persistable[DTOType <: A, EntityType <: C] {
// this would be implemented somewhere else
private def doPersist(source: EntityType): EntityType = source
def persist(source: DTOType): EntityType = doPersist(source)
}
However, this does not compile either.
2) Why does compilation fail at persist and not at the actual method call (val persisted = Persister().persist(B()))? That should be the first place where the actual type of EntityType and DTOType are known, right?
3) Is there a better way to do what I am trying to achieve? Again, this is not the actual thing I am trying to do, but close enough.
Apologies in advance if this question is ignorant and thanks a lot in advance for your help.
You need to make the conversion available within the trait. You can't pass it in from the outside implicitly because the outside doesn't know that persist secretly requires realPersist which requires an implicit conversion. This all fails even without considering JPersistable.
You can for example add
implicit def view: DTOType => EntityType
as a method in the trait and it will then compile. (You can drop realPersist then also.)
Then you need a way to get that view set. You can
case class Persister()(implicit val view: B => D) extends Persistable[B,D]
and then you're all good. (The implicit val satisfies the implicit def of the trait.)
But now you have bigger problems: your Java interface signature doesn't match your Scala signature. The equivalent Scala is
trait JPersistable[T <: Object] { def persist(t: T): T }
See how persist takes and returns the same type? And see how it does not in your Scala class? That's not going to work, nor should it! So you have to rethink exactly what you're trying to accomplish here. Maybe you just want to make the implicit conversion available--not pass it to the method!--and have Scala apply the implicit conversion for you so that you think you've got a persist that maps from DTOType to EntityType, but you really just have the EntityType to EntityType transform that the Java interface requires.
Edit: for example, here's a working version of what you posted just using standard implicit conversion:
trait JPer[T] { def persist(t: T): T }
class A
case class B() extends A
class C
case class D() extends C
trait Per[Y <: C] extends JPer[Y] {
private def doIt(y: Y): Y = y
def persist(y: Y) = doIt(y)
}
case class Perer() extends Per[D] // "with JPer" wouldn't add anything!
object Maps { implicit def BtoD(b: B): D = D() }
object Test extends App {
import Maps._
val persisted = Perer().persist(B())
}
Pay attention to which types are used where! (Who takes B and who takes D and which direction do you need a conversion?)