Imported From: https://issues.scala-lang.org/browse/SI-6457?orig=1
Reporter: @axel22
Affected Versions: 2.9.1, 2.10.0-M7

@JamesIry said:
ImmutableSetFactory says

  def newBuilder[A]: Builder[A, CC[A]] = new SetBuilder[A, CC[A]](empty[A])

GenericCompanion says

  def empty[A]: CC[A] = newBuilder[A].result

So, um, boom.

@JamesIry said:
The parallel case with ImmutableMapFactory is instructive

scala> object Factory extends scala.collection.generic.ImmutableMapFactory[scala.collection.immutable.HashMap] {}
<console>:7: error: object creation impossible, since method empty in class MapFactory of type [A, B]=> scala.collection.immutable.HashMap[A,B] is not defined
       object Factory extends scala.collection.generic.ImmutableMapFactory[scala.collection.immutable.HashMap] {}
              ^

The reason is that ImmutableSetFactory mixes in GenericCompanion by way of generic.SetFactory by way of GenericSeqCompanion. ImmutableMapFactory does not mix in GenericCompanion.

@JamesIry said:
Given the above analysis I don't think this can be fixed without breaking forward compatibility. So kicking it down the road.

@adriaanm said:
Un-assigning to foster work stealing, as announced in https://groups.google.com/forum/?fromgroups=#!topic/scala-internals/o8WG4plpNkw

@adriaanm said:
Unassigning and rescheduling to M6 as previous deadline was missed.

@retronym said:
Marking as a blocker so we don't forget about this one before 2.11

@adriaanm said:
Should we reschedule for RC1? Rex, could you perhaps have a look in time for M8?

@Ichoran said:
Looking now.

@Ichoran said:
Everyone who implements it successfully actually overrides empty. So I propose we just remove the definition of empty and leave it abstract. Although it is formally possible to define the empty set inside the builder, in practice that's normally awkward. I'll submit a PR if we wish. Ideally you'd deprecate the non-abstractness to give people a chance to change, but I don't think we have that annotation? (It's the lesser-known cousin of deprecatedOverriding.)

@Ichoran said:
Oh, I should add by "everyone" I mean "just the Scala library, but nobody in the giant bundle of jars that Adriaan got off of Maven inherited from ImmutableSetFactory, so it's just us that we have to worry about, probably".

@adriaanm said:
The .001% of Scala users hit by this source incompatibility are likely to forgive us.

@adriaanm said:
scala/scala#3366