Right.. our codebase is already unicode all over the place because otherwise we could not i18n properly. So we basically have py2 unicode-correct code with enco

Right.. our codebase is already unicode all over the place because otherwise we could not i18n properly. So we basically have py2 unicode-correct code with encode/decode in the proper places for interfacing witg stdlib. I didn't consider people might use str and not unicode for text..