- From: Lee Feigenbaum <lee@thefigtrees.net>
- Date: Thu, 11 Oct 2007 00:24:35 -0400
- To: 'RDF Data Access Working Group' <public-rdf-dawg-comments@w3.org>, Jonathan Marsh <jonathan@wso2.com>
In a message of many, many moons ago, Jonathan wrote to this list about the WSDL file that defines the SPARQL protocol, suggesting that we add schemaLocation attributes to our XSD import elements, to aid in processors that use the schemaLocation as a hint for finding the schema (such as processors involved in the WSDL 2.0 test suite). In a recent off-list mail conversation with Jonathan, he noted (w/ permission to post this publicly): """ Not sure I'm able to provide a general best practice guideline, but AIUI, this is the situation: 1) schemaLocation is always treated as a hint, and doesn't preclude acquiring a schema for a namespace in some other way. Thus adding it seems harmless enough. 2) schemaLocation provides benefit for implementations without a full schema resolver. 3) The sparql-protocol-query.wsdl as you publish it and the sparql-protocol-query.wsdl as it appears in our test suite differ slightly (schemaLocation attributes). This is an issue for ongoing maintenance, in which we'd prefer that you were able to publish your wsdl directly into our test suite without any modification whatsoever. 4) Maintenance of the WSDL test suite will lie with the upcoming WS Core WG, so the test suite isn't necessarily frozen in its current state, and could continue to be used by vendors to verify that they support all the features used by sparql-protocol-query.wsdl - a good thing for sparql interop IMO. 5) However, the sparql-protocol-query.wsdl hasn't changed significantly in the last 16 months, so manual maintenance is also tractable. """ Apologetically, we didn't address this issue until now, but given this advice we've decided to accomnodate the suggestion. I've added explicit schemaLocations to all of the DAWG WSDL and XML schemas that import schemas. For the moment, the change is only in the local editors' copies of the files, but it will reach /TR/ space when we next publish, which should be within a few weeks. Jonathan, thanks for the implicit and extended patience, and please let us know if you have further comments or concerns. Lee
Received on Thursday, 11 October 2007 04:24:50 UTC