The fact that character numbers can be declared as UNUSED suggests to
me that the intention is that all character numbers be described.
4.32 also points in that direction:
NOTE - Specific characters are assigned to character classes in four
different ways:
...
d) explicitly, by the document character set (NONSGML).
> The text seems quite vague on this point.
I wouldn't claim my interpretation is the only one possible. (In fact
it is not what the currently released version of SP implements.)
However, I think it is a useful one because it allows users to control
what character numbers are allowed in numeric character references by
choosing what characters they declare as UNUSED in the SGML
declaration. If you are working with a parser that can handle large
character numbers in numeric character references, it is useful to be
able to get the parser to reject numeric character references with
character numbers > 255.
> Has this question been put to Charles or WG8?
Not as far as I know. I'm Cc'ing the comp-std-sgml list to get some
more expert SGML opinion.
James