Wish I had documented my exact frustrations, but it wasn't not being able to choose the encoding...something more of the flavor of:
Most of the time I just trying to shuffle bytes from one place to another, and didn't really care about the contents of the bytes, but I hit numerous bug because of exceptions due to things like "external data wasn't valid utf8" (NOT helpful to throw an exception here), and bugs caused by 'str' vs 'unicode' confusion (made more tricky by lack of static type system).
The one time I did actually need to do some calculation involving unicode I got burned by behavior that didn't match the documentation, because my bullshit package-manager-provided variant of python had some insane compile time option that made it pretend utf16 code units were the same thing as actual utf code points, to which I can only say "ha ha ha...fuck those guys".
Can't speak to python 3, haven't used it, probably never will :)
