Unexpected outputs in `scala3-library_3:3.8.0-RC3` jar

[WojciechMazur]

In 3.8.0-RC2 the scala3-library JAR was empty, in RC3 it contains some .class and .tasty files. Most likely caused by semantics conflicts in #24645

The same problem can found in latest nightly version 3.8.1-RC1-bin-20251208-0bb0433-NIGHTLY

Compiler version

3.8.0-RC3

Expectation

scala3-library_3 arifacts should be empty. Duplicate entries on the classpath would negatively affect compiler and tooling

[hamzaremmal]

It is also in the nightlies... https://repo.scala-lang.org/ui/repos/tree/General/local-maven-nightlies/org/scala-lang/scala3-library_3/3.8.1-RC1-bin-20251208-0bb0433-NIGHTLY/scala3-library_3-3.8.1-RC1-bin-20251208-0bb0433-NIGHTLY.jar

[hamzaremmal]

Because of this PR: #24483. I had to change the classDirectory to make metals happy but it had the effect of adding class files when packaging.

[tgodzik]

Are you sure it was necessary, that PR didn't anyway work with Metals as far as I remember?

[unkarjedy]

JFTR, a related ticket in IntelliJ:
https://youtrack.jetbrains.com/issue/SCL-24759/Scala-3.8-RC3-duplicate-JARs

[unkarjedy]

Also, given this precedent, some high-level assertion is needed for the libraries artefacts contents in some tests

[hamzaremmal]

that PR didn't anyway work with Metals as far as I remember?

It did work.