Re: sparql-protocol.wsdl updated

Kendall Clark wrote:
> On Mon, Mar 21, 2005 at 04:54:24PM -0600, Dan Connolly wrote:
> 
>>On Mon, 2005-03-21 at 15:10 -0500, Kendall Clark wrote:
>>
>>>Les chiens,
>>>
>>>I've updated
>>>     
>>>     http://www.w3.org/2001/sw/DataAccess/proto-wd/sparql-protocol.wsdl
>>>
>>><!-- $Id: sparql-protocol.wsdl,v 1.4 2005/03/21 20:00:34 kclark Exp $
>>>
>>>I consider this to be nearly complete w/r/t the "abstract" portion of
>>>the protocol; that is, the interfaces, their types, operations, and
>>>faults.
>>
>>Cool... I'm interested in WSDL tools that do cool stuff with it...
>>are there WSDL validators and such?
>>
>>Have you tried it out in any tools?
> 
> 
> This is WSDL 2, and I don't know of any such tools for WSDL
> 2. Apparently, though, there is a planned remapping of WSDL 2 back
> into WSDL 1.1 -- but I'm very fuzzy on the details.
> 
> 
>>> The changes include importing the results format, declaring
>>>schema types for "the rdf dataset", for some operation response types
>>>(graph creation & deletion),
>>
>>hmm... graph creation and deletion? 
> 
> 
> They were in the previous version, and Andy and I had some discussion
> of them on list, iirc.
> 
> 
>>we decided which interfaces and operations in Boston...
>>
>>  RESOLVED: that the SPARQL WSDL description shall have 3 interfaces
>>  (SPARQLQuery and SPARQLDiscovery and SPARQLQueryAndDiscovery),
>>  each with one operation
>>
>>  -- http://www.w3.org/2001/sw/DataAccess/ftf5-bos.html#item_03
>>
>>The decision is a little funky, because it depends on a decision
>>that I thought we had made earlier but didn't. Still...
>>I suspect you're more likely to get consensus by sticking
>>to just those three.
> 
> 
> They are trivial to remove, if the WG doesn't support them. One nice
> artifact of using WSDL. Plus no one has to implement them, but
> services (like some of ours) which want to can implement them in a way
> that any client can use them if it is written to look for them.
> 
> Seems like a win all the way around.

I don't think that having one particular set of update operations in the SPARQL 
protocol is necessarily a win.  The particular set proposed is one way of doing 
it but there are choices here as to the relationship of graphs and datasets 
(e.g. operations that are more oriented to aggregators with push publishing of 
graphs) and whether the whole graph operations are the best building blocks.

For your services, isn't the point of a W3C recommendation to get other 
implementers to do update the same way so you can work with their services?  It 
isn't a simple matter of implement-if-you-want.

How about splitting the WSDL into two separate documents, one which is purely 
query and service discovery and one which contains update.  Then  other update 
approaches are compared along side update and not query+service description+update.

	Andy

> 
> Kendall Clark
> 

Received on Tuesday, 29 March 2005 10:11:48 UTC