Imported From: https://issues.scala-lang.org/browse/SI-8330?orig=1
Reporter: @gkossakowski

@magarciaEPFL said:
Just checked, the new backend http://magarciaepfl.github.io/scala/ isn't prone to this bug.

@gkossakowski said (edited on Feb 23, 2014 10:30:12 PM UTC):
Miguel, the old wasn't either until we regressed in 9506d5273b4b10037c202f01e8556076998bd064 2 weeks ago.

Do you keep the new backend up-to-date with master?

@magarciaEPFL said:
The new backend is updated twice a week, as of this writing it's based on master as of four days ago. The webpage http://magarciaepfl.github.io/scala/ gets updated accordingly to announce what that "latest branch" is.

It would help if Jira had separate components for old and new (bytecode emitter, optimizer) as a better means to track those bugs, rather than me pasting notes-to-self of the form "the new backend doesn't exhibit this bug".

@retronym said:
"Notes to self" are better expressed as pull requests with test cases!

@gkossakowski said:
I updated the test case. Something a lot smaller is only needed to trigger this bug:

class A {
  def f: Boolean = {
  	try {
      val x = 12
      ()
	} catch {
		case ex: Exception => null
	}
    true
  }
}

@gkossakowski said:
Minimized even more:

object A {
  def f(b: Boolean): Boolean = {
  	if (b) {
      val x = 12
      ()
	} else null
    true
  }
  def main(args: Array[String]): Unit = f(true)
}

results in VerifyError:

java.lang.VerifyError: (class: A$, method: f signature: (Z)Z) Inconsistent stack height 1 != 0
	at A.main(t8330.scala)

The fix is coming shortly.

@gkossakowski said:
scala/scala#3580