scala – 为什么函数尾不是递归的？

Functions like approximate,which call themselves as their last
action,are called tail recursive.

所以,我试过这个：

object Main extends App {
    implicit val mc = new MyClass(8)
    val ti = new TestImplct
    ti.test
}

class TestImplct {
  def test(implicit mc : MyClass): Unit = {
    println(mc.i)
    mc.i -= 1
    if(mc.i < 0){
      throw new IllegalArgumentException
    }
    test
  }
}

class MyClass(var i : Int)

IDEONE DEMO

但它会生成以下堆栈跟踪

Exception in thread "main" java.lang.IllegalArgumentException
    at TestImplct.test(Main.scala:13)
    at TestImplct.test(Main.scala:15)
    at TestImplct.test(Main.scala:15)
    at TestImplct.test(Main.scala:15)
    at TestImplct.test(Main.scala:15)
    at TestImplct.test(Main.scala:15)
    at TestImplct.test(Main.scala:15)
    at TestImplct.test(Main.scala:15)
    at TestImplct.test(Main.scala:15)

这意味着它为每个递归调用生成一个新的堆栈帧.但最后一步是呼唤自己.什么是错的,如何让它尾递归？

为什么编译器不进行尾调用优化？