I am now moving to the utf8everywhere.org way of doing things. So I'm keeping utf8 as my internal encoding and as output of all my files. When I need to talk to windows, I am going to convert my utf-8 strings into whatever Windows expects.
So what encoding does Windows expect with calls to FindFirstFileW? I am compiling with unicode enabled, so I need to widen my utf-8 so that it matches the encoding that FindFirstFileW uses.