Guillaume Martres

You could also keep the runtime support without using it, just to support old versions.

> This has become less pressing now that we don't mix in forwarders in subclasses Not true anymore since `-Xmixin-force-forwarders` was set to `true` by default for performance reasons, right...

/cc @adriaanm since I think you're the resident lazy-vals-in-scalac expert :).

Crashes due to -Yrangepos validation affect IDE and tooling (e.g. scalameta) users, so it'd make a ton of sense to backport https://github.com/scala/scala/pull/6754

Apparently 2.12 can emit code that causes ambiguous method errors in javac starting with Java 11, this should be fixed by https://github.com/scala/scala/pull/7035. I assume that 2.11 is affected by this...

Why ? It's possible to change the lazy val scheme without breaking binary-compatibility.

Also see https://github.com/lampepfl/dotty/issues/7140 / https://github.com/lampepfl/dotty/pull/6979 for the latest proposal.

I don't remember a specific warning coming up ?

I think the confusion stems from the use of "eventual" in the title which usually means "inevitable" in English but which I assume Adriaan used to mean "possible" because of...

> Update Zinc to understand that jrt:// classpath entries should have a timestamp based on the JDK (maybe $JAVA_HOME/jmods/*.mod? Find out exactly what file backs jrt://) Surely that's not right...