In fact it's so trivial I can do it by eye
. No kidding, charset conversion errors are so frequent that you get used to interpret that numerical codes inside a square (or gibberish characters in the opposite direction).
So maybe I'll learn how to cross-compile, dig into the sources and try to fix it. Or maybe I'll just resort to the eye method. I don't have so much time for IRC these days, let alone IRC on a Kindle.
Anyway, I thought these matters were handled by locale now. 'locale -a' only shows utf locales on my KT and there's is not locale-gen, but I wonder if generating the proper locale and setting the corresponding LANG environment variable would suffice.
(this is now the topic of the thread
)