Named Arguments & Implicits

[m4dc4p]

Fixes a bug where implicit definition is not resolved when using named arguments at call site. (https://issues.scala-lang.org/browse/SI-7809)

Updates README with info on using SBT_OPTS to increase JVM heap size if OOM occurs.

[m4dc4p]

CLA signed.

[m4dc4p]

@SethTisue

[lrytz]

Hi & welcome!

This is going to be more complicated, unfortunately.. Simply removing the argument names leads to unexpected / wrong results.

class Test {
  def m(a: Int) = 0
}

object A extends App {
  implicit class Ext(val f: Test) {
    def m(notMyName: String) = 1
  }
  assert((new Test).m("works") == 1)
  assert((new Test).m(a = "compiles") == 1) // weird, there's no parameter named `a` in the selected method
}

object B extends App {
  implicit class Ext(val f: Test) {
    def m(a: Int, b: Int) = a - b
  }
  assert((new Test).m(b = 5, a = 10) == 5) // ups...
}

[lrytz]

This value is really big, 1536m is good enough. Also, we should mention here again that we recommend https://github.com/paulp/sbt-extras, which sets -Xmx1536m by default. So it could be something like:

Note: If compilation fails with "Out of Memory" errors, set SBT_OPTS=-Xmx1536m, or use sbt-extras (recommended) which defines this setting by default.

Feel free to submit this as a separate PR.

[m4dc4p]

Excellent counter-example, thanks for looking.

[m4dc4p]

@lrytz Can I rebase and force push this PR, or is it better to close and create a new branch?

[SethTisue]

rebasing and force-pushing is totally fine, that's usually how we work

[SethTisue]

closing for inactivity. we can reopen it if activity resumes