Arguably Unicode (UTF-8 and -16) doesn't necessarily make this any easier. Or any variable-length encoding, really. You see halved code points quite frequently, and if not that, then halved " and the like.