Deprecate automatic () insertion

[scabug]

While trying to bridge the gap between tuples and parameter lists, the compiler is a bit too eager to insert 0-Tuples () into argument lists.

This is pretty much never what a developer wants (and we couldn't come up with a single example where the () insertion is “a good thing” in three months), but has been the cause of a lot of embarrassing behavior:

scala> List(1,2,3).toSet()
res0: Boolean = false

import java.text.SimpleDateFormat
val sdf = new SimpleDateFormat("yyyyMMdd-HH0000")
sdf.format()
java.lang.IllegalArgumentException: Cannot format given Object as a Date
  at java.text.DateFormat.format(DateFormat.java:301)
  at java.text.Format.format(Format.java:157)
  ...

scala> class A[T](val x: T)
defined class A

scala> new A
res0: A[Unit] = A@5534578f

Mailing list threads:

https://groups.google.com/forum/#!topic/scala-internals/4RMEZGObPm4
https://groups.google.com/forum/#!topic/scala-user/4yggJc38tIg/discussion
https://groups.google.com/forum/#!topic/scala-debate/zwG8o2YzCWs

[scabug]

Imported From: https://issues.scala-lang.org/browse/SI-8035?orig=1
Reporter: @soc
Affected Versions: 2.11.0-M7

[scabug]

@retronym said:
I think we should consider this as part of a holistic plan to deprecate autotupling in cases when the expected type was not TupleN.

[scabug]

@SethTisue said:
Man. It'd be super great if this finally got addressed.

[scabug]

@soc said:
PR: scala/scala#3260

[scabug]

@Atry said:

import scala.util.control.TailCalls._
def gotWarning: TailRec[Unit] = done()
def noWarning: TailRec[Unit] = done(())

I don't know why scalac warning on done() now.

[scabug]

@soc said:
done takes an argument, but you are not providing one in the first example.