> Or we find an interoperable way to transport/encode the control
> characters (agree on entities or char references or PIs).
I would very much prefer to do this than to allow those naked codes to appear
in text. I support the idea of finding a way to encode such data, rather than
include it per se (as per Derek's suggested change in focus).
Numeric character references () are essentially the same as the literal
data (once parsed the distinction is lost) so I would not support their use.
PI's, while being one mechanism, are application-specific, so are probably
not ideal.
That leaves us with entities. Perhaps something along the lines of creating a
"virtual" enitity set in the &Unnnn; space? This was suggested in the ERCS
days...
"an XML 1.1 processor may interpret entity references beginning with the
letter 'U', followed by 4 hexadecimal characters as representing an
entity holding the representation of the Unicode Scalar equivalent of
the number."
This would provide a standard naming scheme for entities representing code
points, but leave the exact resolved value undefined. No value is necessary
anyway, as the entity reference provides all the needed information.
