For instance, ß, which in UTF8 is 0xC39F, turns into 0xDF.
How do I get it to stop doing this? For the short term, I have inserted calls to utf8::encode from the UTF8 module before I write the string out, to convert them back to UTF8, however I'd really like to fix the thing so it doesn't convert the strings to US-ASCII in the first place.
From what I can tell, Perl does something to mark strings as UTF8 somehow, so I should maybe do something along those lines? I don't know.