Tail call optimisation

When I have code like this:

int bar() {
  // Something
}

int foo() {
  return bar();
}

I understand that Pike 8.0 and 8.1 optimise the call to bar() away and turn
it into a jump?

What do Pike 8.0 and Pike 8.1 currently do with this:

void bar() {
  // Something
}

void foo() {
  bar();
}

Does it optimise the call to bar() out similarly but substituting it with
a jump?

Or do we need to write it explicitly as:
void foo() {
  return bar();
}

to get that optimisation?
-- 
Stephen.