At 12:55 PM +0100 8/27/03, David Carlisle wrote:
the problem is what to XML parsers do if they don't read the specified
dtd but do see an entity. Most seem to generate an error, even though
the XML spec says that that is not a well-formedness error.
Mozilla's XSLT engine for example never gets a chance to be conformant
or not in this situation as the XML parser barfs on the input document
before XSLT starts. (If there was a cast iron clause in some spec
somewhere that could be used to persuade mozilla that this was bad
behaviour that would be useful, but it seems to be a grey underspecified
area to me)
My interpretation is that the XPath data model simply does not allow
unresolved entity references, and therefore an XSLT processor or any
other thing built on top of the XPath data model must use a parser
that resolves entity references.
--
Elliotte Rusty Harold
elharo(_at_)metalab(_dot_)unc(_dot_)edu
Processing XML with Java (Addison-Wesley, 2002)
http://www.cafeconleche.org/books/xmljava
http://www.amazon.com/exec/obidos/ISBN%3D0201771861/cafeaulaitA
XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list