- From: Savas Parastatidis <Savas.Parastatidis@newcastle.ac.uk>
- Date: Tue, 17 Jun 2003 23:56:11 +0100
- To: "Steve Graham" <sggraham@us.ibm.com>
- Cc: "David Snelling" <d.snelling@fle.fujitsu.com>, "Jim Webber" <jim.webber@arjuna.com>, <ksankar@cisco.com>, "Paul Watson" <Paul.Watson@newcastle.ac.uk>, <public-ws-desc-state@w3c.org>, "Steve Tuecke" <tuecke@mcs.anl.gov>, "Umit Yalcinalp" <umit.yalcinalp@oracle.com>
I get the feeling that my position is now understood but not accepted, which is fair enough (it seems that I am the minority anyway). Please allow me to reiterate my position through a concrete example. I agree that a solution that meets all the requirements would be preferable (still have issues with the accessibility/permissions requirements which I believe are an orthogonal issue but I won't start a second front:-). I do believe that the serviceData approach cannot be implemented by the WSDL specification as is because of the use of specific operations. Here is why... In order for the requirements of Service Data to be met, a number of operations are introduced as part of the GridService interface, namely setServiceData, findServiceData. At least, that's my understanding of the proposed approach. These operations are defined in terms of WSDL (I am assuming the move from serviceData to attributes): <interface name="MandatoryAttributeRelatedOperations"> <operation name="setAttribute"> <input message="SetAttributeInputMessage"/> <output message="SetAttributeOutputMessage"/> </operation> <operation name="getAttribute"> ... </operation> <operation name="findAttribute"> ... </operation> </interface> This was a translation (serviceData->attribute) from the OGSI specification. Now, the above interface, perhaps enriched with more operations, is going to meet the requirements that are set for attributes. How is this going to be described in the WSDL specification? Firstly, it uses WSDL already and secondly, WSDL is defined using the Infoset component model. How are the above operations going to be described using that Infoset model? Perhaps I should wait until a more concrete proposal is available and then restart the discussion, if needed. It may be the case that my current worries are not justified. .savas.
Received on Tuesday, 17 June 2003 18:56:31 UTC