summaryrefslogtreecommitdiff
path: root/test/files/pos/ticket2251.scala
diff options
context:
space:
mode:
authorPaul Phillips <paulp@improving.org>2012-08-10 16:19:19 -0700
committerPaul Phillips <paulp@improving.org>2012-08-11 08:29:22 -0700
commitdb46c71e8830639bc79e6363332a06642fd3d8cc (patch)
tree52b5457a53bac5119522611cbbc4d40da1a1f76b /test/files/pos/ticket2251.scala
parent48d9fb7307fb6519fe786a7d9be97996c5812fb7 (diff)
downloadscala-db46c71e8830639bc79e6363332a06642fd3d8cc.tar.gz
scala-db46c71e8830639bc79e6363332a06642fd3d8cc.tar.bz2
scala-db46c71e8830639bc79e6363332a06642fd3d8cc.zip
Improvement for SI-2251, failure to lub f-bounds.
After a great struggle, I realized that the major reason that code like this still doesn't compile: List(Stream(), List()) is that we were poisoning the computed lub in mergePrefixAndArgs by throwing in Any when the max recursion depth was reached. I modified it to return NoType instead, which allowed me to teach lublist to recognize what has happened and fall back to a weaker type, one which does not contain recursive bounds. This enables the lubbing process to complete. The most elusive lub, defeated. Notice also that the refinement members are correctly parameterized on Nothing, rather than on Any as has often been the case. scala> List(Stream(), List()) res0: List[scala.collection.immutable.LinearSeq[Nothing] with scala.collection.AbstractSeq[Nothing]{def companion: scala.collection.generic.GenericCompanion[scala.collection.immutable.LinearSeq with scala.collection.AbstractSeq]; def reverse: scala.collection.immutable.LinearSeq[Nothing] with scala.collection.AbstractSeq[Nothing]{def companion: scala.collection.generic.GenericCompanion[scala.collection.immutable.LinearSeq with scala.collection.AbstractSeq]; def reverse: scala.collection.immutable.LinearSeq[Nothing] with scala.collection.AbstractSeq[Nothing]{def reverse: scala.collection.immutable.LinearSeq[Nothing] with scala.collection.AbstractSeq[Nothing]; def dropRight(n: Int): scala.collection.immutable.LinearSeq[Nothing] with scala.collection.AbstractSeq[Nothing]; def takeRight(n: ...
Diffstat (limited to 'test/files/pos/ticket2251.scala')
-rw-r--r--test/files/pos/ticket2251.scala14
1 files changed, 14 insertions, 0 deletions
diff --git a/test/files/pos/ticket2251.scala b/test/files/pos/ticket2251.scala
index b3afee4ea9..c220e85350 100644
--- a/test/files/pos/ticket2251.scala
+++ b/test/files/pos/ticket2251.scala
@@ -22,4 +22,18 @@ lub of List(D, C) is B[_2] forSome { type _2 >: D with C{} <: B[_1] forSome { ty
// should be: B[X] forSome {type X <: B[X]} -- can this be done automatically? for now, just detect f-bounded polymorphism and fall back to more coarse approximation
val data: List[A] = List(new C, new D)
+
+ val data2 = List(new C, new D)
+
+ val data3: List[B[X] forSome { type X <: B[_ <: A] }] = List(new C, new D)
+
+ // Not yet --
+ // val data4: List[B[X] forSome { type X <: B[X] }] = List(new C, new D)
+ // <console>:7: error: type mismatch;
+ // found : List[B[_ >: D with C <: B[_ >: D with C <: A]]]
+ // required: List[B[X] forSome { type X <: B[X] }]
+ // val data4: List[B[X] forSome { type X <: B[X] }] = List(new C, new D)
+
+ // works
+ val data5 = List[B[X] forSome { type X <: B[X] }](new C, new D)
}