When I was playing with the ports4dotclear system (see the previous post about it), I had to fix a number of Unicode handling errors in both Python and Perl. For those of you who do not know what is Unicode, it's the technology which allows me to write this kind of stuff: «æ€¶ŧ←»® or «你好世界».
I was forced to use Unicode because, well, XML-RPC communications are in XML and thus require the content to be encoded in UTF-8 (a Unicode encoding/decoding charset).
All this mess leads me to this conclusion : Unicode sucks in Python 2.x ! (and Fedora i18n has still some strange artifacts)