The MIME type ought to take care of NOTATION issues; that is,
the MIME type should be mapped to the appropriate NOTATION if
need be. The arbitrary data need not be parsed; you could insert it
the rendering stage, after the parse, although if the content
turned out to be SGML you'd have to start a new parse (as though
it were a SUBDOC) and insert the rendered contents of that
second parse.
It is true that if arbitrary characters are dumped into an
element (even a CDATA one) and </ occurs there will be a parsing error.
One could get around this by inserting CDATA marked section
delimiters around EMBEDded data:
<p>foo
<![ CDATA [
this is some </test text
]]>
end of para.</p>
This does not work if the CDATA marked section occurs in a
CDATA element, I learn by experiment.
-- Terry Allen (terry@ora.com) Editor, Digital Media Group O'Reilly & Associates, Inc. Sebastopol, Calif., 95472