I am not sending this to the `ietf' list.
I would suggest that any text formatting be handled similar to
X-window Motif strings, which are a C-type XmString.
The suggestion here is to use OSF's Compound Strings for English and
non-English text in Internet email. (OSF happens to use C for their
implementation of Motif, and `XmString' is the name of the type they
use for Compound Strings.)
Compound Strings are rather like ISO 2022 (and hence, the MIT X
Consortium's Compound Text), since they allow intermixing of several
coded character sets. Each Compound String consists of a number of
tag-length-value segments. The one-byte tag identifies the type of
segment, the two-byte length gives the size of the segment, and the
value is the segment's data itself. One possible type for the tag is
the charset type. This indicates that the value of the segment
contains a character encoding identifier, which affects all subsequent
segments until another charset is specified. Another tag type is
text, which, of course, indicates that this segment is actual text.
One of the problems with using Compound Strings in Internet email, is
that we cannot use them now and expect many people to be able to read
them. Of course, if we all had software that could render Compound
Strings, and could convert to and from the 7-bit representation
required for SMTP, there would be no problem. But how do we get from
here to there? We need to think about the migration, and Keld's
mnemonics are one way to ease the transition.
Note that this is also true of other character encodings such as
Latin-1, ISO 10646, and so on.