I like Python 3's unicode handling but I agree that this seems strange. It is because people expect to see "characters" from these interfaces after treating the

I like Python 3's unicode handling but I agree that this seems strange. It is because people expect to see "characters" from these interfaces after treating them as ASCII-only for so long. If Python 3 had insisted on real purity with bytes objects I think it would have died a long time ago. Which is sad.