RE: No specification for querystrings and fragment identifiers

> From:	Dan Connolly []
> 	<mydoc>foo &quot; bar</mydoc>
> &quot; doesn't have the status of &lt; and &amp; in XML, because
> it doesn't need to; you can just write:
	We were talking about parameter values (in particular
	href).  I'm also using XML to some extent as a surrogate
	for the SGML ISO standard and because of HTML's migration
	to XHTML.  Real HTML, where this started, doesn't allow one
	to add entities, or at least I'd be surprised to find
	many browsers that supported it.

> >         RCDATA appears not to be supported, meaning that
> >         re-declaring %URI as RCDATA is no longer an option.
	I partially confused the issues here because I 
	mistakenly thought that RCDATA was a valid type
	for a parameter.  It appears, that, at least for
	XML, parameter CDATA is like SGML content RCDATA.

> again, no need:
> 	<mydoc aRef=";def"/>
	This is not the case that I was considering.  The
	case was that of actually getting &quot; as a value
	for the parameter.  From the other reply and section
	3.3 of the XML specification (for the moment I shall
	assume that this does not extend the SGML specification,
	to which I only have second hand access) you can get
	this value into a parameter by using &amp;uot;

	(NB the original problem didn't require this, it was
	just to demonstrate an apparent lack of generality.)

	Incidentally, section 2.4 is still confusing, as it
	says "To allow attribue values to contain both single
	and double quotes, the apstrophe or single-quote
	character (') may be represented by '&apos;'....

	The presence of this statement tends to imply a special
	exception for ' and ", but not for &, whereas, once
	section 3.3 is pointed out, it may just be mentioning
	particular cases of the general case (although, without
	&, not a self consistent subset).  If the XML spec is
	ever re-drafted, I think that this should either be taken
	out, have a reference to section 3.3 added, or replaced
	by a statement that attribute CDATA is treated differently
	from CDATA sections, together with a reference to section 3.3.

> >         Looks to me as though there is a serious conflict between
> >         what is implemented and what is consistent with the
> >         standards which are authorative in specific areas.
> How so?
	If section 3.3 of the XML specification describes the
	same behaviour as the ISO standard for SGML does, with
	respect to CDATA attributes, then there isn't a problem.

Received on Tuesday, 23 November 1999 14:05:39 UTC