Re: New work item for XML group ? (Re: Comments on 31 March spec)
Let's be realistic. There *will* be cheap 8-bit-internal XML systems.
There *will* be 16-bit-internal systems. There may be 32-bit-internal
systems. There may even be non-canonical-internal (e.g., shift-JIS)
I doubt that XML will prescribe that it is illegal to have an XML system
that cannot represent (and display, else why bother representing) all
the Eastern character sets.
A question we need to address is: "What should an XML system do when
presented with a numeric character reference for a character for which
it has no internal representation?"
A second question we should be asking is: "Should XML permit numeric
character references using non-SGML (i.e., undefined in the document
character set parameter) character numbers?" This determines just how
one should write the document character set parameter for XML's SGML
declaration. Note: If they are permitted, how is the system to know
how to handle (e.g., display) them? How will it "know what the
A related question that SGML generally has to think about, but which
impacts XML is: "What character is referenced by a numeric character
reference using a defined-in-the-document-character-set character
number which is defined to be a "character" represented in a base
character set by a code point that the base character set specifies
to be undefined?"