What's bad news? that you have to manually add the encoding attribute to the generated xml dat?
when not using xml dats you run into issues with non standard characters anyway since they differ from codepage to codepage and you have to toggle the oem/ansi conversion for archives...in other words....don't use non standard chars
With utf-8 encoded dats (and you need to specify the encoding) you should at least be able to have datfile xmls which can be read in the whole world without caring about different characters.
When I'm opening your datfile is see a superscripted 3 character and no 'sword' / 'cross' character as in your posts. So...before we can test something, we first have to define...what do you want to see