The irony here is that the dependency hell within pure Scala libraries has improved tremendously and is not really a problem anymore, yet we need to deal with m

The irony here is that the dependency hell within pure Scala libraries has improved tremendously and is not really a problem anymore, yet we need to deal with many binary compatibility issues due to transitive Java dependencies. Things that regularly cause issues, for instance when assembling JARs that get shipped to a Spark cluster at work: Guava, Jackson, Kryo, Log4j, Hadoop's transitive dependencies (like Netty) ... not Scala's fault.