>It mentions that HTTPResponse should do unicode -> DEFAULT_ENCODING. I >think that HTTPRequest should do backward translation. Or am I missing >something why it shouldn't?
Absolutely correct. That's why I asked others to join in, because I am bound to have forgotten some parts ;-) I added that to the page and added a special case for GET parameters, as those don't carry any information on used charset. Actually I think we need to decide on what to do when decoding won't give a proper unicode string. The "clean" thing to do would be to throw an exception, an "unclean, but practical" way could be to have a "fall back encoding" to use when the primary one failes (like using utf-8 as the primary and latin1 as the fallback that is only used if the primary gives an exception - reasoning being that in that case you at least get the string, even though it might be encoded wrong). bye, Georg
