Rainer: I made a comment earlier about how we should specify character set in terms of Unicode and UTF-8. You may want to reference RFC 2277, which provides guidance on "IETF Policy on Character Sets and Languages". Basically it says three things:
- must support UTF-8 character set/encoding - must provide capability for identifying the language - must have section "Internationalization considerations" I think internationalization should be part of -protocol because support of UTF-8 already implies it. So, we need to add the last 2 things. Identification of a language is tricky. I think we should not require it. For example, we need to be able to fire an authentication message in English which contains a username in Japanese. So, how do we identify the language such that one knows which language applies to which part? I'd suggest we don't. Maybe we should just define an optional multi-valued structured content parameter like "lang" which would optionally list all languages used by the message. We should support RFC 1766 language tags. Thanks, Anton.