- From: Michael Rys <mrys@microsoft.com>
- Date: Wed, 14 Jan 2004 13:50:23 -0800
- To: "David Carlisle" <davidc@nag.co.uk>, <xquery@comcast.net>
- Cc: <public-qt-comments@w3.org>
> It is the suggestion that sometimes you need to do that and sometimes > you don't that makes this problematic. [Michael Rys] I agree that you always should entitize XQueries in the trivial embedding... > > Also, embedding XQuery queries by wrapping them with CDATA sections > clearly > > doesn't work when a query contains the sequence ]]> > > You can't put the sequence ]]> in an XML file anyway it's not just that > you can't wrap a CDATA section around it, that sequence isn't legal > anywhere (except ending a cdata section. [Michael Rys] This is clearly incorrect. An XML document clearly can contain ]]>. > An embedding of xquery into XML that is safe and reversible is > replace each of < & and > by < & > place resulting string in > XML element. > Extract element content (not its linearisation) and parse as Xquery. > > That is essentially what you do for Xpath anyway. The danger is the > suggestion that unless there are < being used as < (rather than element > constructors) the Xquery can be directly placed in the XML. > > David > > > -- > http://www.dcarlisle.demon.co.uk/matthew >
Received on Wednesday, 14 January 2004 16:50:34 UTC