OGSI Session at GGF9: GWSDL to WSDL 1.1 mapping Date: 10/07/03 Minutes: Mario Antonioletti (mario@epcc.ed.ac.uk) Actions Taken at the Meeting: o Sam Meder to look insertion of the GGF copyright notice into the GWSLD to WSDL 1.1 mapping document. o David Snelling will check that the namespace descriptions at the beginning of the GWSDL to WSDL 1.1 document are still accurate. o The following folks have volunteered to carefully read through the document when it goes into final call: - Peter Vanderbilt - Sam Meder - William Lee o Sam Meder to re-run the GWSDL2WSDL program on the exanmples in the document and check the output is consistent with what is there. Also to check that the UML is still correct. o David to circulate to the OGSI mailing list that discussion/process on OGSI trackers is about to begin. --- David Snelling opened the meeting. Encouraged folks to move forward to the front as at some point he might dispense with the microphone. GGF IP Policy covered. Main topic for discussion was the GWSDL to WSDL 1.1 transformation document. Required to use the OGSI specification - need to get from GWSDL to WSDL 1.1 format so that it can be used by current tooling. Going to try to approach final call with this document as soon as possible. This is one of the missing bits in the formal chain. Based on first draft want to get approval on the changes. - Use extension rather than inheritance wording. - Need to insert GGF copyright (Sam Mendel will look at inserting GGF copyright) - Deleted a comment made by Steve Graham. - Word changes. - Need to check namespaces to see that they are still accurate (David will do this) Savas Paratstatidis: comment on whether there is any tooling for WSDL 1.2 on the first page. Document has used a specific draft of the WSDL 1.2 spec for the normative parts. Bits and pieces of the original OGSI spec were going to be put in appendices - this is no longer deemed to be necessary. References will be inserted instead. Taken out another comment by Steve Graham. Want people to go through the document, esp section 3.1, when the document goes to final call - three names ... Action: folks that will carefully read through the document when it goes to final call - Pete Vanderbilt - Sam Meder - William Lee Deleted a comment by Steve Graham and another comment made by David Snelling. Comment made on the recursion dealt with in point (d) on page 5 of the document. Point made about comment "Note: this may be incomplete" at the top of page 5 - David was not sure what "this" referred to in the context of the paragraph. Can only think of 2 possibilities: - something in the process of writing the document which no longer applies. - something to do with the types. First hunch is to delete it. The reverse process is trivial - going from WSDL 1.1 to GWSDL - want to make sure that are people are happy with this. There is nothing to tool GWSDL today so there is not much point in doing this. It can easily be done manually. No dissent. One question from David: there is an assumption going in the forward direction by leaving the GWSDL bits it cannot understand that it will not do anything with it. Is this a safe assumption? Savas concurred. That is where the normative text ends - should non-normative text stay or be removed? David is happy to leave it in. It talks of a hypoethical tool. Sam Meder: it might be a good idea to put it in an appendix. David agrees. Added some text to take out one of Steve Graham's comment. Action: Sam Mender to re-run the program to make sure that the output is consistent with what is in the document and to check the UML in the document is still correct. A normative reference to the OGSI spec was also missing that will be pasted in by David. David asks for questions: no questions came from the audience. Used the remainder of the time through the tracking items that are outstanding in gridforge. OGSI purposefully stayed quiet about a second version of the OGSI spec to encourage adoption of the first version. This has happened. It might be safe now to start talking about a second version. Refresh our memory about the stuff that is still in the tracker system. Trackers: Tracker 95: Specify semantics of Service data and operations: there was a long discussion about the meaning of SDEs. Peter: OGSI decided that there would be no decorations - there were two kinds of state - service data state and the real state. Pete would be happy for this tracker to go away. Savas: will not make a comment - waiting to a WS vs GS discussion to be held later on in the day. David: do not have to decide now - can do this at a later telcon, Will append decisions made here to the trackers for discussion in the telcons. Conclusion: Possibly drop. Tracker 96: People in DAIS have raised issues about this. Nobody in DAIS could expound on how they got round the restriction. David is inclined to leave this one as it is - possibly a drop. Savas: would use another word as opposed to polymorphic. Malcolm: happy with the use of polymorphism. Savas: polymorphism implies inheritance ... Malcolm: no it does not ... Discussion will be taken off line .... Conclusion: possible drop. Pete: How do you find bug 34? Have to go back to the bugzilla page and look it up there. It might be worthwhile cutting and pasting the original bug into the tracker ... Tracker 97: has anyone had problems finding their portType references? no reply. Conclusion: possibly drop as out of band. Tracker 98: has anyone had problems finding service data or implement re-direction. Conclusion: this is still interesting so will leave as is. Tracker 99: Pete: Two comments: 1. Who needs it? Someone might not need it now but it may be needed later. 2. Was this not a delivery type where you could specify an operation that you had to call that was raised by S. Graham? (1) a negative answer does not mean it will be taken out - looking for a positive answer, (2) this was another issue which has been tracked elsewhere. First Conclusion: this was changed to drop. Pete: steve thought was quite important and should be involved in the discussion. Tracker 100: Asked if anyone knew the status of WS-timestamp? WS-Timestamp is used in WS-Security. It is published in OASIS. WS-Security currently in review. Sam: are timestamps used anywhere? David: yes they are. Conclusion: this will be left on the discussion list. Tracker 101: Notification currently works on SDEs and not on destruction. Might have to create an pseudo SDE to work within the framework but would require additional infrastructure to make services more heavy weight .... Conclusion: suggest that this is dropped. Tracker 103: Is this not included in SOAP? David asked Savas. Savas asked for clarification about a unique identifier. Savas thought that this was orthogonal to OGSI. Conclusion: possible drop as orthogonal to OGSI. Tracker 104: Will not open. GT3 already supports XPath queries to findService. Tracker 106: David cannot remember the context for this. Best left until later. Tracker 280: Came from the last GGF meeting that arose from a DAIS meeting. Pete put it in. It might be interesting to try to put in a portType that support an iterator model. Malcolm: this kind of programming should be discouraged as it might encourage folks to mis-use this kind of interface. If DAIS are interested in this then they are wrong. David - I think this should be dropped as it does not belong in the OGSI layer and should probably go in the OGSA layer if at all. It is probably a programming issue. Conclusion: put this as a possible drop. Remaining trackers mainly deal with a comment sweep - best left to a face-to-face. Savas: has a couple of trackers to submit that have arised from the OGSI primer. 1. Inter-operability with the WSDL description of the GSR as the GSR is extensible - how is it done then? 2. Who is responsible for assigning the GSH of a service - the handle resolver or the factory that creates the service. The GSR format is currently very open ended. There is WS-Addressing - is this something that could be used? Need to clarify the end point modelling. Sastry: it is not an open-spec yet so it cannot be used, It's not even been submitted to OASIS yet. Wording of Sava' proposed trackers: 1. GSR Format - are there emerging standards? Two possible places: WS-Addressing WSDL service to be used as an endpoint reference. Sava: WS-Addressing defines something that is so simple that OGSI could easily replicate this functionality. David: GT3 already uses similar constructs. There are a lot of inter-operability issues to address within the community. The OGSI spec is ambiguous as to whether the factory or the handle resolver creates the handle. Sam: In GT3 it is the factory Savas: have to check what was put in the primer.... Peter: the ogsi spec was silent on this.... Sastry: the issue is not so much who creates it but how it is managed.... Savas: once a new service instance is created by a factory.... (missed the rest). Another point from tracker: 3. A factory can return an empty locator for the service created. The explanation text may have to go in the primer. Dave Berry: point about compliance testing ... David: GGF is about inter-operability and not compliance. Malcolm: point as to how do you establish inter-operability if there is not a mechanism to do this .... David: Need to define levels of inter-operability .... Comment: service types do not include life time attributes - so life time attributes should be added to service data elements.... David: have to go and think to even start answering. Need to make some types as optionals in the schema description. Meeting was drawn to a close.