scala.tools.nsc.javac.JavaParsers can't parse JDK14 records

[xuwei-k]

reproduction steps

https://openjdk.java.net/jeps/359

https://github.com/scala/scala/blob/a20345d2380dfc63ff0aa41d4ab97b538a9f74a7/src/compiler/scala/tools/nsc/javac/JavaParsers.scala#L874

$ java --version
openjdk 14 2020-03-17
OpenJDK Runtime Environment (build 14+36-1461)
OpenJDK 64-Bit Server VM (build 14+36-1461, mixed mode, sharing)

build.sbt

javacOptions ++= Seq("--enable-preview", "--release", "14")
scalaVersion := "2.13.1"

A.java

package aaa;

public record A(int x, int y) { }

B.scala

package bbb

class B {
  def x = new aaa.A(3, 4)
}

sbt version 1.3.8

problem

A.java:3:15: illegal start of type declaration
[error] public record A(int x, int y) { }
[error]               ^
[error] one error found
[error] (Compile / compileIncremental) Compilation failed

expectation

no error.

workaround

// build.sbt
compileOrder := CompileOrder.JavaThenScala

[mghildiy]

Is this taken up?

[SethTisue]

@mghildiy nearly certain nobody is working on this

[mkurz]

records will be out of preview starting with Java 16, to be released in March: https://openjdk.java.net/jeps/395
Would be nice to see support in Scala soon.

[SethTisue]

This would make an excellent project for an interested volunteer.

[harpocrates]

I'd be interested in picking this up.

Just to confirm: does scala.tools.nsc.javac.JavaParsers need to properly detect and error on all invalid Java code (eg. an abstract record), or is it OK to sometimes to accept Java code which javac will later reject? As I understand it, the critical bit is that if the Java source does compile, the output of JavaParsers must match the types/members javac produces.

[som-snytt]

@harpocrates yes iirc only what is required for interop.

I happen to be reading Plato atm. Socrates wrote nothing. Presumably, Harpocrates spoke nothing. Appreciate the midday chuckle.