Kefas,
you have written:
I tried UTF-8 export to send an e-mail that contained several scattered unicode codepoints from the full 16-bit range from oooo to ffff from XP+Word to the university's Linux/Mozilla/OpenOffice/Kmail, enabled UTF-8 support. With very disappointing results.
For UTF-8 (or any other encoding except ISO 646 IRV (aka ASCII)) to survive the transport via e-mail (RFC 2821), it must be tagged and "transfer-encoded", according to RFC 2045, and RFC 2047. For examples, cf. <http://www.systems.uni-konstanz.de/EMAIL/FAQ-SMTP.php#74> (in German). It is the e-mail clients' responsibilty to do this tagging and encoding (on the sending side), and the corresponding interpretation and decoding (on the receiving side).
You have not mentioned, which e-mail client program you have used, how it was configured, nor what the result looked like. Hence, the cause of your "very disappointing results" cannot be derived (nor even guessed at).
1. Do I expect too much assuming that UTF-8 just recodes the full 16-range in 8-bit but that text-programs with UTF-8 enabled should be able to reconstruct the full 16-bit range (as far as used)?
The Unicode range is much more than 16 bit (you need 21 bits per character, but all 21 bit values are not used). UTF-8 encodes every single character in 1 through 4 bytes; cf. <http://www.unicode.org/faq/utf_bom.html> for more details. I do not understand what you mean by "recon- struct", but I guess your question is answered in the cited WWW page.
Best wishes, Otto Stolz
