Minutes of SCA-Bindings TC Conference Call
August 19, 2020
Simon H chaired, Tom Rutt took minutes.
1. Introduction
1.1 Roll:
Meeting Attendees
Tom Rutt Fujitsu Limited* Group Member
Bryan Aupperle IBM Group Member
David Booz IBM Group Member
Mike Edwards IBM Group Member
ant elder IBM Group Member
Simon Holdsworth IBM Group Member
Martin Chapman Oracle Corporation Group Member
Anish Karmarkar Oracle Corporation Group Member
Ashok Malhotra Oracle Corporation Group Member
Plamen Pavlov SAP AG* Group Member
Laurent Domenech TIBCO Software Inc. Group Member
Eric Johnson TIBCO Software Inc. Group Member
1.1 Agenda bashing
Proposed agenda:
1. Introduction
2. Approval of the minutes from 29 July:
http://www.oasis-open.org/committees/download.php/38836/SCA%20Bindings%20minutes%202010-07-29.doc
3. Actions
20090211-4 [General] Write up HTTP binding use cases
Completed
20100729-1 [Simon Holdsworth] Query TC Admin on status of Test Suite PR request
completed
20100729-2 [Eric Johnson] Develop use cases for HTTP binding
completed
4. New Issues
http://www.osoa.org/jira/browse/BINDINGS-134
The JMS Binding schema for wireFormat.jmsDefault and operationSelector.jmsDefault is invalid
http://www.osoa.org/jira/browse/BINDINGS-135
JMS Binding nativeOperation/selectedOperation inconsistency
http://www.osoa.org/jira/browse/BINDINGS-136
JMS Binding operationProperties headers element should be optional
http://www.osoa.org/jira/browse/BINDINGS-137
JMS Binding attribute 'jndiName' is required in the schema but described as optional in spec text
5. JMS Binding testcase progress
Ant is making progress on these; target for test cases document and test cases ready for review 26th August
6. HTTP binding use cases
As per emails on mailing list including:
http://lists.oasis-open.org/archives/sca-bindings/201008/msg00018.html
7. Impacts of 1.2 changes on bindings
Discuss how to assess impacts of 1.2 changes on the bindings, including any 1.2 education needed
8. Open issue discussion
http://www.osoa.org/jira/browse/BINDINGS-60
JMS Default wire format insufficient to cover real world usage
Raiser: Mike Edwards, owner: Simon Holdsworth
Priority: 3 (deferred)
Status: outline proposal in issue
9. AOB
No comments agenda accepted.
2. Approval of minutes from last meeting
http://www.oasis-open.org/committees/download.php/38836/SCA%20Bindings%20minutes%202010-07-29.doc
No objection, minutes approved
3. Action Items
20090211-4 [General] Write up HTTP binding use cases
Completed
20100729-1 [Simon Holdsworth] Query TC Admin on status of Test Suite PR request
completed
20100729-2 [Eric Johnson] Develop use cases for HTTP binding
completed
4. New Issues
http://www.osoa.org/jira/browse/BINDINGS-134
The JMS Binding schema for wireFormat.jmsDefault and operationSelector.jmsDefault is invalid
Mike: references to substitution groups get an error when referring to base format. The extended production needs an element and a type. The extension needs a ref to the global element.
Mike moves to open bindings 134, Bryan Seconded.
No objection, bindings 134 opened.
Given proposal is incorrect, defer resolution to complete proposal over email that passes validation.
http://www.osoa.org/jira/browse/BINDINGS-135
JMS Binding nativeOperation/selectedOperation inconsistency
Simon: an old issue Bindings 44 was supposed to have fixed this, however it was not done properly. We need a new issue to track this fix.
Mike moves to open bindings 135, Bryan seconded.
No objection, bindings 135 opened
Simon: proposal is straightforward.
Mike moves to resolve bindings 135 with proposal in Jira, Bryan seconded.
No objection, bindings 135 resolved with proposal in Jira
http://www.osoa.org/jira/browse/BINDINGS-136
JMS Binding operationProperties headers element should be optional
Simon: schema omits the use attribute, however it works with header out.
Bryan: rev 1 headers has a min occurs = 0.
Mike: That means we can close this issue with no change to schema.
Mike moves to close bindings 136 with no action, Bryan seconded.
No objecti0n, bindings 136 closed with no changes to schema.
Action: fix schema in spec to agree with the machine readable schema xsd file for Bindings 136.
http://www.osoa.org/jira/browse/BINDINGS-137
JMS Binding attribute 'jndiName' is required in the schema but described as optional in spec text
Mike: rev 1 has the correct schema. We should vote to close this with no change.
Mike moves to close bindings 137 with no changes, Bryan seconded.
No objection, bindings 137 closed with no change.
Action: Fix schema in spec to agree with the xsd file for bindings 137.
5. JMS Binding testcase progress.
Ant: Most of them are passing in Tuscany now. I still have to write up the test cases.
Simon: so we should have them written up in the next week or two.
6. HTTP binding use cases
http://lists.oasis-open.org/archives/sca-bindings/201008/msg00018.html
Eric reviewed the first used case in email from the above link.
Anish: on first use case, http binding in wsdl 1 and wsdl 2.0 had this use case in mind. Putting aside support for such bindings, the current bindings.ws should be able to support this use case.
The wsdl 2.0 binding has useful information, however there is not a whole lot of implementation uptake on the wsdl 2 and http binding.
Eric: Assembly rejected supporting wsdl 2.0. It is not readily feasible to use WSDL 2.0 to support the http binding. JAX RS lets you annotate a java class mapping methods to URLs and defining notions of resources and resource collections.
Anish: does Jax RS support WADL.
Eric: not sure it supports WADL.
Simon: Jax RS focuses on rest access to a java class.
Anish: I bring up the wsdl 2 and http binding work as something for us to learn from.
MartinC: i just wanted to say that seems like this is cart before the horse. Until (or if) we relax wsdl requirements working on bindings is premature and wee need to have an architectural discussion in assembly.
MartinC: as anish alluded the worlds of wsdl and rest/http are hard to reconcile but would have to be done at the assembly level
Eric proceeded to explain the second scenario in the use case.
Eric: This use case is for an atom feed. You periodically execute an operation asking if there is something new, if there is it returns the new feed info.
Eric: you can have a separate wsdl operation for each atom feed.
Eric: for this last case it should be thought of as a different kind of interface. You would want to define an interface.atom, with a well know mapping to wsdl using constraints.
Eric: the last scenario is the inverse of the first one, you have to do translation to come up with html.
anish: mapping of "REST" to WSDL (or other way around) me things is bound to fail (in terms of adoption) -- at the very least let's not call it REST
Simon: Bryan and I responded with comments on these scenarios. One of the key questions is the mapping in http binding to a wsdl interface the best way forward.
Eric: there is a notion of xml response, that might want to be mapped to json, xhtml or other form. Are you saying such a mapping has to be defined by assembly first?
Simon: if you are starting from wsdl interface you specify how you get from message parts to http url and content.
Simon: this is addressed in Jax RS, one possible approach is to do as annotations to wsdl interface.
Eric: what is assembly's role for clarifying this.
Simon: assembly might define new interface types.
Mike: how would this look as different implementation types. That is, how to map a java interface in a particular way to make sense in a rest setting. What has to be said for a bpel process to offer a rest interface, and also how to act as a client for a rest style servant. These two cases, java on one side and bpel on the other are to be addressed. Json is part of the data binding problem. whatever we come up with we have to be able to deal with across languages. Whatever statements are made about binding.http have to be able to deal with restful interfaces in several environments.
Bryan: the binding has to reflect acceptable formats, formats provided, how parameters get mapped. Some of these are data binding concepts, but we need to separate this from the interface definition discussion. Wsdl may not be the natural mapping for such things as atom feeds. We need to understand how a bpel process consumes a feed.
Simon: the intention to having these use cases written is to allow members to determine if there is justification for defining an html binding for sca. If we do feel that these use cases are sufficiently compelling, we can look at the document we currently have to satisfy these use cases. We can then decide if such a modification is appropriate for sca bindings.
Simon: I will schedule agenda item for next meeting to discuss the http binding.
Meeting adjourned at 1:00 PM EDT.