scala/scala

A relaxed chat room about all things Scala (2 & 3 both). Beginner questions welcome. https://scala-lang.org/conduct/ applies

Luis Miguel Mejía Suárez

@BalmungSan

@agaro1121 does:

lazy val otherModule=
  (project in file("foo"))
    .dependsOn(moduleWithResources % "compile->compile;test->test;provided->provided")

work?

Anthony Garo

@agaro1121

Let me try that

raymond95

@raymond95:minds.com

[m]

agaro1121 (Anthony Garo): are you interested

som-snytt

@som-snytt

@Sciss that is a bug, as it should warn if there is a value Void in scope, not the java type. Probably it's confused by internal representation of java things as companions. It needs a place for Void.TYPE.

som-snytt

@som-snytt

I think the nowarn works on the enclosing def, @nowarn def f = ....

Also charming notation:

scala> def f() = (s match { case "$Void" => }): @annotation.nowarn
def f(): Unit @scala.annotation.nowarn

Hanns Holger Rutz

@Sciss

thanks

raymond95

@raymond95:minds.com

[m]

hardlianotion (hardlianotion): hi

Sciss (Hanns Holger Rutz) how are you

zeroexcuses

@zeroexcuses

Machine A, B are two x86_64 linux machines with identical software and identical hardware. How complicated is it, while a server written in scala is running on machine A, to migrate it to machine B? (Assume open tcp connections, but no open files).

Hanns Holger Rutz

@Sciss

in current sbt which is very picky about version mismatches, is it possible to tell it to shut up while I work with locally pub lished snapshots? e.g.

incompatible:
[error]
[error]     * de.sciss:audiofile_2.13:2.4.0 (pvp) is selected over 2.4.0-SNAPSHOT
[error]         +- de.sciss:mellite-core_2.13:3.7.0-SNAPSHOT          (depends on 2.4.0)
[error]         +- de.sciss:scalacollider_2.13:2.7.1                  (depends on 2.4.0)
[error]         +- de.sciss:sonogramoverview_2.13:2.3.0               (depends on 2.4.0)
[error]         +- de.sciss:lucre-synth_2.13:4.9.0-SNAPSHOT           (depends on 2.4.0-SNAPSHOT)

Guillaume Martres

@smarter

I don't know, but if you're publishing the dependencies yourself you can also change their versionScheme to make sbt happy

Hanns Holger Rutz

@Sciss

oof

I prefer to temporarily set tyhe policy in the dependent project to treat x.y.z-SNAPSHOT same as x.y.* ?

Hanns Holger Rutz

@Sciss

tells me to look at evictionErrorLevel but sbt docs don't reveal anything about it

ThisBuild / evictionErrorLevel := Level.Info does the trick

ahad1234

@ahad1234

Hey All!
i am trying to mock one of the scala objects to test some functions,
can anyone of you tell me the possibilities with it?

With some tries I get this exception:
java.lang.NoSuchMethodError: org.mockito.internal.invocation.InterceptedInvocation.getRealMethod()Lorg/mockito/internal/invocation/RealMethod;

Matthias Berndt

@mberndt123

OK, so why does this not compile?

trait Foo {
  type A
  def f(ta: T[A] => Unit): Unit = ()
}

sealed trait T[A]
case object I extends T[Int]
case object S extends T[String]

trait Bar { self: Foo =>
  f {
    case I => println("Int")
    case S => println("String")
  }
}

[error] /tmp/test/src/main/scala/mberndt123/Main.scala:12:10: pattern type is incompatible with expected type;
[error]  found   : I.type
[error]  required: T[Bar.this.A]
[error]     case I => println("Int")
[error]          ^
[error] /tmp/test/src/main/scala/mberndt123/Main.scala:13:10: pattern type is incompatible with expected type;
[error]  found   : S.type
[error]  required: T[Bar.this.A]
[error]     case S => println("String")
[error]          ^
[error] two errors found

It says the pattern type is incompatible, but there's actually no way for it to know that.

Guillaume Martres

@smarter

it compiles in scala 3.

Matthias Berndt

@mberndt123

seems like I'll need to migrate then...

Matthias Berndt

@mberndt123

And of course I've got a bazillion dependencies with macros.

Seth Tisue

@SethTisue

@zeroexcuses not sure what you mean by "migrate". also it seems like a JVM question rather than a Scala question per se, if I understand correctly?

maybe even an OS-level question rather than a JVM question? I don't think a TCP connection can "migrate" like that, OS-wise? (if my guess by what you mean by "migrate" is correct)

Matthias Berndt

@mberndt123

So I have this trait:

private[wernicke] trait Foo[G[_]]:
  def apply[L, R]: PartialFunction[(G[L], G[R]), Result[G, L, R, ?]]

And I thought I'd be able to use Scala 3's polymorphic function literals:

val f: Foo[G] = [L, R] => { case (Bla, Blurb) => ??? }

Guillaume Martres

@smarter

Yeah that's not supported yet

Matthias Berndt

@mberndt123

And I get

[error] -- Error: /home/mbt/newp/src/test/scala/NewTest.scala:141:13 
[error] 141 |      [L, R] => {
[error]     |             ^
[error]     |Implementation restriction: polymorphic function literals must have a value parameter

that's disappointing :-(

Is that being worked on?

or is there some workaround?

Guillaume Martres

@smarter

Polymorphic SAM types support is something I'd like to work on but haven't found the time so far

Same for parameterless polymorphic functions

(you'd need both here)

Matthias Berndt

@mberndt123

I see…

Matthias Berndt

@mberndt123

That would be useful for what I'm working on right now.

I can work around it, but it's pretty ugly

type inference for PartialFunction also doesn't seem to work all that great, I need to annotate the type to make it work.

Matthias Berndt

@mberndt123

So here's how I work around the implementation restriction:

sealed trait T[A]
object T:
  case object I extends T[Int]
  case object S extends T[String]

trait Foo[G[_]]:
  def apply[L, R]: PartialFunction[(G[L], G[R]), Unit]

class FooPartiallyApplied[G[_]]:
  def apply(f: [L, R] => Unit => PartialFunction[(G[L], G[R]), Unit]) =
    new Foo:
      def apply[L, R] = f(())

object Foo:
  def apply[G[_]]: FooPartiallyApplied[G] = new FooPartiallyApplied[G]

val x =
  Foo[T] { [L, R] => (_: Unit) => ({
      case (T.I, T.S) => println("Int-String")
    }): PartialFunction[(T[L], T[R]), Unit]
  }

it's… not good

So I need this (_: Unit) thing because it wants a value parameter, and I need the Foo.applymethod to work around the fact that SAM types are not supported

and I need : PartialFunction[(T[L], T[R]), Unit] to convince the type checker

but I think at this point it's better to just give up on the polymorphic lambda syntax for now and just write an anonymous class instead. That's probably easier to understand.

Leif Warner

@LeifW

Is there a JRE-only Docker image available for Java 17 people like to use? The Oracle OpenJDK and Eclipse Temurin ones seem to be full JDK only.

raymond95

@raymond95:minds.com

[m]

LeifW (Leif Warner): yds

Yes LeifW (Leif Warner)

jodersky

@jodersky

Hi! I have a question regarding some recent-ish documentation changes to the Quotes API. Recently, this comment in Quotes.scala came to my attention, specifically the part that talks about symbol.tree.tpe being an antipattern:

Warning: avoid using this method in macros.
[...]
Anti-pattern: The following code is an anti-pattern:
 symbol.tree.tpe
It should be replaced by the following code:
 tp.memberType(symbol)

I don't understand the suggested alternative. What is tp in this example, and where would it come from?

I have to admit that I have actually used this pattern quite frequently in the past. The most prominent case being a macro which inspects the types of a method's parameters obtained via a symbol. It looks something like this:

for (paramsSyms <- method.paramSymss) {
  for (paramSym <- paramsSyms) {
    paramSym.tree.asInstanceOf[ValDef].tpt.tpe.asType match { // <---this is the antipattern, how could this be cleaned up?
      case '[t] =>
        // do something based on the param type...
    }
  }
}

This has always felt as a bit of a contortion, and now that I see that it is indeed considered an antipattern, I wonder what alternative there could be.

raymond95

@raymond95:minds.com

[m]

jodersky (jodersky): okay DM me I will help you out

Where communities thrive

People

Repo info

Activity