what encoding is the X clipboard text in?

There is a slight misfeature in Mule XEmacs when working under X. When I type some Cyrillic text in XEmacs, then select it with mouse and paste it to xterm, I get various clearly wrong results. I understand the issues about e.g. mixed English-Russian-Chinese, which cannot be handled by eight-bit xterms). I think at least the simplest/user-defined case should be somehow worked out. Sometimes when I'm pasting text I get it in iso8859-5, internal Mule encoding. I want it to be in koi8-r (or other Cyrillic encoding of my choice). For that I need something like (set-clipboard-ccl-programs 'ccl-encode-koi8-r-font 'ccl-decode-koi8-r-font) (so that pastes from e.g. xterm to XEmacs would work too). When I try to paste from xterm (which AFAIU copies text to clipboard as a single-byte bytes in koi8-r) to XEmacs, I get the following there: %/1─┬koi8-rPRIWET The "PRIWET" part corresponds to the appropriate koi8-r characters with stripped eighth bit and flipped case (whoa!). I have a slight feeling that I've seen a function for that, but I can't find it anywhere. Sometimes also I get the Cyrillic string pasted to XTerm as a series of tilde characters, just like it happens when Mule can't find fonts for used encoding. I can't yet figure out how to exactly reproduce that. I'd be glad to test any patches for that matter. --alexm