Curses! ... WIDE curses.

Over the last couple weeks, I've been discovering that the business of writing wide-character programs that use curses is almost completely undocumented. I'm trying to figure this out because there are apparently some people who seem to like utf8, and I want them to stop bothering me :-). I was wondering if anyone out there has experience getting ncurses to play nicely with Unicode (or, I suppose, other wide character) locales, and if so, if there's any standard documentation that I could be pointed at. The ncurses manpages have some general information, but they're missing key details (for instance, what are the precise semantics of getcchar/setcchar, and what character encoding should I be passing into them?), and it looks like I'll probably have to dig through the library code to get real answers. *sigh*...

[UPDATE] the library code reveals that my worst fears are realized. getcchar/setcchar operate on whole strings. But, you see, they only let you pass in a single attribute/color value. So despite the fact that you're supposedly manipulating attributes combined with text (like chtype), you're actually unable to do anything more complicated than set the attributes of a whole string at once -- or carefully feed your data to curses one character at a time, generating pointless single-character strings to do so! Ewwwwwwwwwwwwwwwwwwwwwwwwwwwwwww.

[UPDATE 2] In the previous update, I somehow missed the prominent usages of CCHARW_MAX in getcchar() and setcchar(). setcchar errors out if there are more than CCHARW_MAX characters in the string, and getcchar never copies more than CCHARW_MAX characters from its input cchar_t to the output. (there seems to be some funny business going on with the number of wchar_t values in the string vs wcslen, but I could well just be confused)