SRT's Questions which are to be found at http://lists.w3.org/Archives/Public/public-ws-chor/2004Apr/0051.html Document under discussion forms basis of Working Draft but includes line numbers. This is to be found at: http://lists.w3.org/Archives/Public/www-archive/2004Apr/0004.html http://lists.w3.org/Archives/Public/www-archive/2004Apr/att-0004/cdl_v1-editors-apr03-2004-pdf.pdf Question 1: Lines 217-219 does this mean no WSDL is required? Does it mean same WSDL is required? Does it mean different WSDLs are possible? Answer: We need to figure out what this may mean. The text is abstract enough to incorporate any direction that we choose to take. The current situation would suggest that WSDL2.0 is required. (see line 250) Question 2: Lines 269-270: Does this mean WSDL2.0 or earlier or both? Answer: See section 1.4 Question 3: Lines 295, 299, 503 and more I'm sure: Do we need the term "business processes"? Can we just have "processes" since WS-CDL may be used in areas in which the term business process has less relevance. Answer: Yes okay. Question 4: Line 321-322: "Semantics allow the creation of descriptions that can record the semantic definitions of almost every single component in the model" If it is "almost" then what is excluded? Answer: We need to verify what almost means. We need to check why it is not included in some constructs. Need to file an issue and follow it through Question 5: Line 441, 459: Dumb question from me. What is the relationship between 'role name="ncname"' and 'role type="qname"'. Maybe I don't understand what ncname is. I presume qname is qualified name relative to some schema. Answer: ncname is not a qname - so not qualified. Begs the question why one over the other - that needs clarification. namespace names are qnames and thing in the CDL that are ncnames define names in the CDL namespace which is why target name space is in a CDL Question 6: Section 2.3.3 (Relationships): Would I be correct in thinking about a relationship as a static connection between processes? Answer: If a process is an abstraction of some named behavior then this correct at least for one case. Kind of makes sense. Question 7: Back to Semantics above: What would semantics be and how would the look? Answer: Semantics are descriptive text which may be structured in RDF to enable other tools to make some assertions about a CDL. Question 88888888: Line 449 and 547-551: Does this mean that a channel is a web service and so they are coupled? Answer: This is an abstraction layer. Today it is bound to WSDL2.0. But should this need to be changed it can be without any adverse impact. Question 9: Line 534-535: Is there any reason not to have numbers other than 1 and unlimited? Answer: This is really question for our invited experts. Question 10: Line 573: Is the the effective correlation id in this example? Is this how correlation will be performed relative to channels? Answer: Yes. Question 11: Line 579-580: "that can influence the observable behavior" In what way can a WS-CDL document influence behavior since it is a description? Does this only apply to generation in which case I can see such a relationship? Answer: Yes the latter questionis right. Question 12: Lines 602-607: Do these variables describe the relationship between messages? i.e. A is composed of specific parts of B Is this correlation? Answer: No it is not intended to be correlation. It's sort of alpha renaming on the one hand. It could be used to guide skeleton generations for variabled on invoke and receive in BPEL. Question 13: Line 624: "a Channel Variable could contain information such as URL to which a message should be sent" Is this a "could"? Answer: Depends on channel type and so is a could. Question 14: Line 674: What is a silent-action used for? Alas no example to illustrate. Answer: The silent-action represents something that is not observable. Perhaps our invited experts might help us with a use case. Question 15: Lines 739-745: Given the definition previously why is "part" valid? Is this an XQuery thing? Answer: This is a mistake. "Part" should be removed. Question 16: Line 748: Is "prescribes" the right word here? Answer: Describes is better Question 17: Lines 872-878: How do I guard with more than one variable? Is there any notion of existential qualifier associated with this so that I can have a guard that simply says: If there exists an x, y and a z then .... Answer: You can use as many variables as you wish. Extension functions on XPath in the WS-CDL document include getVariable. Question 18: Section 2.4.8.1 Exception Block. How is the guard in an Exception Block work unit related to the non-exceptional work units in the enclosing choreography? Are they always the same? Are they always non-overlapping? Are they unrelated? Is it user defined? What is the consequence of un-restriction? Answer: It probably doesn't make sense for the same guard to be used. Raises some model checking issues because there may be a need to validate guards against one another. Question 19: Lines 1094-1099: What if there are no matches for an guarded work unit in an exception block? Answer: See above? The exception block might well be caught higher up. No explicit final catch all is defined in CDL. Question 20: Line 1177: Activity-Notation+ This allows 1 or more not two or more. Answer: Yes it should be two or more. You can express this in the schema however. Question 21: Lines 1211-1212: Is it possible to model a one way receive? Answer: An interaction is a pair. Question 22: Lines 1271-1283: Does this mean that some higher order messaging protocols allow states in the CDL to be inferred rather than captured explicitly? (i.e. if I know that when I send a message from A to B that B gets the message then I can infer something about the state of B relative to the receiving of that message.) Answer: No you cannot infer. It does not guarantee that B read the message as opposed to just received the message. What is needed is guarantee of understanding of the message that was about. Question 23: Lines 1312-1313: If align and initiateChoreography are not present what is the default behavior (true or false)? Answer: Align is defaulted to false. Question 24: Lines 1355-1359: Default behavior again? Answer: See above. Question 25: Lines 1583-1596: Not sure why you use the terms "this" and "free" and what they might mean? Is it to indicate that the fromRole is the "this" and the toRole is "free based on being the initiator (i.e. Directionality?) Answer: "this" is a variable is the aliased name of the free variable. "free" is a variable that is free and so is yet to be bound.