Sean just added unicode to banana and jelly. 1. Will this run on 1.5.2? 2. You had better encode the data before writing it - as far as I can see from the patch you do not. I suggest UTF-8. 3. As glyph pointed out, do we even need unicode support in Banana? Keeping it in jelly (u'foo' --> ['unicode', u'foo'.encode("UTF8")] would be the transform) would work just as well, without requiring people who implement banana to deal with it.