prelude: |
  str = "a" * 1000
  re = Regexp.new(str)

benchmark:
  dup: re.dup