Some overdue improvements in repl completion, w...

Some overdue improvements in repl completion, which has been largely
awol since the pickler format change. Where possible, completion is now
done using the compiler's internal model rather than reflection. Many
handy things now work which did not before, such as wildcard imports
causing all imported identifiers to henceforth be completable. Note also
that there is a verbosity counter now, so hitting tab twice may yield
more results than hitting it once.

scala> import java.util.concurrent.atomic._
import java.util.concurrent.atomic._

scala> Atomic<tab><tab>

AtomicBoolean                 AtomicInteger
AtomicIntegerArray            AtomicIntegerFieldUpdater
AtomicLong                    AtomicLongArray
[etc]

Review by community.

1 files changed, 1 insertions, 1 deletions

diff --git a/src/compiler/scala/tools/nsc/interpreter/Parsed.scala b/src/compiler/scala/tools/nsc/interpreter/Parsed.scala
index b130396cc6..810116d0cf 100644
--- a/src/compiler/scala/tools/nsc/interpreter/Parsed.scala
+++ b/src/compiler/scala/tools/nsc/interpreter/Parsed.scala
@@ -15,7 +15,7 @@ class Parsed private (
   val cursor: Int,
   val delimited: Char => Boolean
 ) extends Delimited {
-  def isEmpty       = buffer == ""
+  def isEmpty       = args.isEmpty
   def isUnqualified = args.size == 1
   def isQualified   = args.size > 1
   def isAtStart     = cursor <= 0