Reviewing the latest spec we have recognized that the fault handling that occures if an greement could not be created (currently 1 fault) is not sufficient for the protocol. There are several issues related to this: 1. currently only the fault ResourceUnknownFault is defined. That fault indicates that the in the EPR referenced AgreementFactory (instance) is not known to the server. 2. taking that in mind the createAgreement method should also define a ResourceUnavailableFault indicating that the AgreementFactory instance is known but it experiences problems at the time. 3. There should be a general CreateAgreementError that indicates that the creation of an agreement failed. There was a ContinuingFaultError defined for this, however it looks to me that the naming of the error produced some problems in the past and it was therefore removed. Perhaps it is feasible to reintroduce this general error but rename it. (e.g. AgreementCreationError) 4. It would be convenient to have some general error codes defined for things that can go wrong when creating agreements. I would suggest the following: <xs:complexType name="AgreementCreationErrorCodes"> <xs:sequence> <xs:element name="GeneralError" type="xs:int" fixed="0"/> <xs:element name="TemplateUnknownError" type="xs:int" fixed="1"/> <xs:element name="InvalidAgreementOfferError" type="xs:int" fixed="2"/> <xs:element name="CreationConstraintsViolationError" type="xs:int" fixed="3"/> </xs:sequence> </xs:complexType> Even if the public comment periode is over I would suggest having at least one separate exception in the schema that can be thrown if an agreement can't be created. Regards, Oliver Waeldrich ----------------------------------------------------------------------- Oliver Wäldrich Fraunhofer-Institute for Algorithms and Scientific Computing (SCAI) Schloss Birlinghoven, D-53754 Sankt Augustin, Germany Tel: +49 2241 14 2830; Fax: +49 2241 14 42830 CoreGRID Network of Excellence www.coregrid.net Institute on Resource Management and Scheduling www.coregrid.net/irms