Scala Buggabot

@viktorklang said: So the "regression" part is nonsense?

@paulp said: I don't understand how there can not be a bug.

@gkossakowski said: Well, we are getting very close to binary compatibility here. What happens is that we generate different byte-code due to changes in specialization related to AnyRef specialization (that...

@gkossakowski said: Paul: Do you understand the exact issue here?

@paulp said: I'm not completely sure except to be pretty sure it's our bug. It has to do with the thing being parameterized as BoxedUnit turning up as Object one...

@gkossakowski said: Well, ok, I'll have another look tomorrow. First I need to minimize this by not depending on Function1 interface which is huge. I'll try to come up with...

@paulp said: There are "official regressions" and there are "practical regressions". Our guarantees in this area are slim to nonexistent, and would be fairly meaningless if they existed, but this...

@SethTisue said: Agree. If javac chokes it's a pretty safe assumption that other tools will choke too and we should be thankful to javac for the early heads up.

@gkossakowski said: Ok, I spent more time on it and we indeed generate wrong signatures for method (that do not agree with generic signatures). I created a separate ticket to...

Imported From: https://issues.scala-lang.org/browse/SI-10092?orig=1 Reporter: @martijnhoekstra Affected Versions: 2.12.0, 2.12.1 See #10057