In article <20060513.235134.74684306.kayama@st.rim.or.jp>, Akihiro KAYAMAwrites: kayama> I'm also using Quixote-2 for my Japanese(UTF-8) application. Here is kayama> a kludge I'm using to set default encoding without modifying global kayama> site module: kayama> kayama> import sys kayama> reload(sys) kayama> sys.setdefaultencoding("utf-8") kayama> kayama> It can't be appropriate, but it works perfectly for getting rid of kayama> annoying 'ascii' codec error. In addition, today I tested the application on Python-2.4.2 + Quixote-2.4 and found that I could get rid of sys.setdefaultencoding(). When I wrote it last year, both Python-2.3 and Quixote-2.0 had mysterious Unicode related problems such as using % operator so I gave up and decided to change default encoding simply. Improvement in Unicode support of current version of both softwares seems to resolve such problems. (CHANGES told me it was done at Quixote-2.2) Perhaps, implicit conversion between unicode and str is evil, or default encoding should not be site global. In article <6e9196d20605131024q51c78338jaafbf776ec9d8c6f@mail.gmail.com>, "Mike Orr" writes: sluggoster> You're not supposed to reload builtin modules. Exactly. I'm happy I can do what I am supposed to at last. Thanks for everyone who contribute these excelent products. -- kayama