I'm not sure how DC++ protocol is supposed to handle filenames and chat messages, but currently it looks like the single-byte characters are sent to network as-is. In particular, russian text seems sent to network in Windows-1251 character set (on Windows, that is), but the Linux GUI displays them as ISO8859-1 (I think). Same (I think) applies to file names: file names on my disk can be in either character set I choose (for example, some use UTF-8 on disk, some like me use single-byte character sets like ISO8859-1, Windows-1251, KOI8-R and so on).
If that's true and DC++ protocol doesn't know the word "unicode", at least a setting for local file names character set, and remote file names/chat character sets (preferably separate, e.g. chat in Windows-1251 because most DC++ users use Windows, and local filenames in, say, UTF-8 because most Linux distributions like UTF-8 nowadays).