Thanks. I will do that.

Upon further reflection, I don't think I can get what I want. Since strings are always characters, no module can tell for certain what it should do with one containing sequences of code points that would be valid utf-8: those sequences can be either multiple latin-1 characters, or an individual character encoded as utf-8. Doing the right thing for knowledgeable users will probably mangle the data of everyone else even more.

Does that sound like enlightenment, or confusion?