RE: No specification for querystrings and fragment identifiers

> From:	JOrendorff@ixl.com [SMTP:JOrendorff@ixl.com]
> 
> What I mean to say is that XML CDATA attributes
> don't behave the way you seem to be saying they
> behave.  Unless I misunderstood you.
> 
	My current understanding, at the end of this
	thread, is that all entity references are expanded
	(and expanded only once++, i.e. the text in 
	section 2.4 doesn't imply that &uot; gets
	turned into " - the fact that it might is one reason 
	why I think it is unsafe) in CDATA attributes in XML.

	My understanding of SGML is that RCDATA content
	(not attributes, as I was incorrect in assuming that
	you could have them) also have all entity references 
	expanded.

	That seems to me to give equivalent treatment between
	XML CDATA attributes and SGML RCDATA content.

	The one question that this leaves is whether SGML
	CDATA attributes have all entity references expanded.  The
	HTML 4.01 specification assumes they do, but relies
	on the ISO SGML standard as its authority.  I only
	have a book describing the standard, and it rather
	glosses over the semantics of CDATA attributes.

	++ but not excluding recursive expansion of complete
	nested entity references.

Received on Tuesday, 23 November 1999 14:55:48 UTC