Scala Buggabot

@retronym said: https://github.com/scala/scala/pull/2986

@adriaanm said: Part of the problem is existentialTransform's use of subst instead of substSym: ```scala def doSubst(info: Type) = info.substSym(rawSyms, typeParams) // don't use subst, as it won't update pre.sym[args]...

@adriaanm said: the type symbols are still out of whack after this, but at least now we select an abstract type rather than a class symbol off of the new...

@adriaanm said (edited on Feb 11, 2014 10:14:56 PM UTC): https://github.com/scala/scala/pull/3510

@adriaanm said: Vaguely related problem: ```scala scala> class Test { | def c: x.X forSome { val x : {type X

@paulp said (edited on Feb 12, 2014 7:29:55 AM UTC): lazy vals are really an innocent victim, their mistake is using a var in the implementation. Here is another, free...

@paulp said: That compiles if it's private\[this\], which is how you can tell it's the correspondence between getter and field. With vars it's even more booched thanks to the setter....

@adriaanm said: I debugged it to subtyping for existentials being broken. ```scala val underlying1 = underlying.instantiateTypeParams(quantified, tvars) // fuse subst quantified -> quantifiedFresh -> tvars ``` does not go into...

@paulp said: You mean "substitution" not "subtyping" right? Not that subtyping for existentials isn't broken.

@adriaanm said: subtyping is broken because substitution is -- you can't tell, because existentialTransform transitively chases all bounds so it doesn't matter, but that's not the right fix (as that...