In message "Re: Some text that may be useful for the update of RFC 2376",
Chris Lilley wrote...
Yes. Though it is not clear what a non-XML-aware transcodes would do when
swizzling an XML document between encodings. Since it can't use entities or
NCRs, what does it use for characters that do not fall within the
repertoire of te encoding it is converting to? Question marks?
I believe that most trascoding of XML documents will be from legacy
encodings to Unicode, since every XML processors can handle Unicode.
If this is the case, we do not need character references or CDATA
sections.
It should always be an error of some kind if the charset parameter
does not agree with the encoding attribute (or other Appendix F
mechanism). Only given that constraint is it useful to make the
charset parameter significant.
I agree, but then given that constraint the charset parameter is
superfluous since it adds no new information. However, I might be prepared
to conceed that it is not too harmful as long as it is constrained tosay
what the XML encoding says, and for it to be an error for these to differ.
I do not understand why the difference between the charset parameter
and the encoding PI is harmful. The arguement against the charset
parameter is its omission. If it is omitted, inconsistency will not arise.
If it is provided, it is authoritative.
----
MURATA Makoto muraw3c(_at_)attglobal(_dot_)net