- From: Norman Walsh <ndw@nwalsh.com>
- Date: Thu, 17 Jul 2008 10:46:51 -0400
- To: public-xml-processing-model-comments@w3.org
- Message-ID: <m2d4lcsh44.fsf@nwalsh.com>
/ "Alex Milowski" <alex@milowski.org> was heard to say: | We had this discussion on one of our telcons quite awhile ago and | the WG was notably against allowing this. In fact, I originally | wrote the step to allow this and the consensus was it was a bad | idea. I think we're going to have to do something about the p:xquery step. The observation that XQuery documents don't have to be namespace well-formed is a show-stopper for our current solution, I think. I can think of three workarounds that *I think* would work: Option 1: Use XInclude, we do nothing: <p:xinclude> <p:input port="source"> <p:inline> <c:query> <xi:include href="document.xqy" parse="text"/> </c:query> </p:inline> </p:input> </p:xinclude> Option 2: Add a step that does this: <p:wrap-text wrapper="c:query" href="document.xqy"> Or, I suppose, we could modify p:document so that it had this behavior with some optional attribute: <p:document href="document.xqy" parse="text" wrapper="c:query"/> That might be the easiest thing to do. Option 3: Allow the query document on p:xquery to come *either* from the source port or from an href option: <p:xquery href="document.xqy"> <p:input port="query"> <p:empty/> </p:input> ... </p:xquery> None of these strikes me as ideal... Be seeing you, norm -- Norman Walsh <ndw@nwalsh.com> | Not everyone can live upstream. http://nwalsh.com/ |
Received on Thursday, 17 July 2008 14:47:35 UTC