- From: Lynn, James (HP Software) <james.lynn@hp.com>
- Date: Thu, 12 Jul 2007 10:29:42 -0400
- To: "John Arwe" <johnarwe@us.ibm.com>, <public-sml@w3.org>
- Message-ID: <5A5CC5E87DE62148845CC96C8868900E0B3B1F11@ataexc02.americas.cpqcorp.net>
John, I agree with your proposed steps below, other comments inline. I would like to propose that we solicit comments from within our WG for submission to the TAG, the goal being to ensure that our use of the terms fits their definition and would hope their goal would be the same. I don't think we want this to become a major undertaking but think it is the role of the WG's to provide feedback to the AG's. Regards, Jim ________________________________ From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On Behalf Of John Arwe Sent: Thursday, July 12, 2007 10:00 AM To: public-sml@w3.org Subject: RE: [w3c_sml][Bug 4651] Definition of 'consumer' needs clarification I am concerned that the note may have been mis-read, may be self-inconsistent, or both. My evidence, pasted from the ref'd Note: [Definition: A producer is an agent that creates text.] [Definition: A consumer is an agent that consumes text.] [easily missed caption under the first block diagram, bold mine] The diagram shows a simplification where the producer and consumer are relationships from agent rather than separate entities. If I were to stick with the definitions as stated above, and ignore the "simplified" diagram, they are necessary but not sufficient wrt what we have in the SML specs for producer/consumer, because there is no notion of understanding. An XSLT processor could consume SML/SMLIF text yet it has no concept of a model or references. Section 1.1.1 uses the term 'understands' (would be nice if it were conciesly defined somewhere) in a way that suggests 'able to process', at least to me. Section 1.1.3 uses the term 'partial understanding' in a way that I think might cover your XSLT example. [Definition: An Act of Production is the creation of text.] [Definition: An Act of Consumption is the processing of text of a language.] "ok". I still see nothing like SML/SMLIF's implications that their producers/consumers "understand" in any way the model(s) being processed. I would suggest that both the TAG and we need to better define what we think should constitute "understanding". By the definitions we are discussing the XSLT processor "understands" the model, at least partially. Therefore it is an SML/SML-IF consumer. Does this violate this WG's intent? [Definition: A Language consists of a set of text, any constraints on the information, and the mapping between texts and information.] [Definition: Information is the result of processing, manipulating and organizing a text in a way that adds to the knowledge of the consumer.]. (Ednote: cribbed from wikipedia). [Definition: The information set of a language is the set of results of processing, manipulating and organizing each of the texts in a language.] Building on the definitions above, I think I could find ways to define SML (and SMLIF) producers/consumers consistent with the above; SML and SMLIF are examplary languages. SML maps texts like sml:uri to information (about a model). SMLIF contains additional mappings for texts like smlif:definitions to the information that the XML infoset content of smlif:definitions consists of model definition documents. In terms of how we could choose to use this information in our specs, one way would be: 1. Copy the relevant definitions into our spec(s), noting their origin precisely and that the ref'd origin is a WIP. 2. After some time has passed, e.g. in preparation for Last Call, re-check the status of the Note. If by then it has become normative/stable, we can choose either to reconcile its then-current content with our definitions, or cut the cord to the original source and carry them independently. 3. If it is still a Note, so we cut the cord above, at a later time when the Note transitions to a stable state and after SML/SMLIF are Recommendations, our descendants/followers can re-visit the "cut the cord" decision either in Errata or in a later version etc. There are probably other ways to enable us to continue to make progress in parallel, suggestions welcome. Best Regards, John Street address: 2455 South Road, Poughkeepsie, NY USA 12601 Voice: 1+845-435-9470 Fax: 1+845-432-9787 "Wilson, Kirk D" <Kirk.Wilson@ca.com> Sent by: public-sml-request@w3.org 07/11/2007 09:37 AM To <public-sml@w3.org> cc Subject RE: [w3c_sml][Bug 4651] Definition of 'consumer' needs clarification Just a note of clarification. The consumer and producer relationships as defined in the TAG Note were not between consumer and producer. Granted that relationship exists in some sense. The consumer relationship as defined in the TAG Note is between the Agent and something called the Act of Consumption. (Similarly for producer and the Act of Production.) Kirk Wilson, Ph.D. Research Staff Member CA Labs 603 823-7146 ________________________________ From: Lynn, James (Software Escalations) [mailto:james.lynn@hp.com] Sent: Wednesday, July 11, 2007 9:16 AM To: Wilson, Kirk D; public-sml@w3.org Subject: RE: [w3c_sml][Bug 4651] Definition of 'consumer' needs clarification I agree that these are better viewed as roles and perhaps this should be brought to the TAG's attention. I don't necessarily have a problem with the relationship concept, certainly there is a relationship between a consumer and producer. But at the same time, we don't want to imply that a producer cannot exist without a consumer. I would like to see comments from the rest of the group as to whether or not we should engage the TAG or just decide autonomously. James Lynn W3C Service Modeling Language WG HP Software 610 277 1896 ________________________________ From: Wilson, Kirk D [mailto:Kirk.Wilson@ca.com] Sent: Wednesday, July 11, 2007 9:10 AM To: Lynn, James (Software Escalations); public-sml@w3.org Subject: RE: [w3c_sml][Bug 4651] Definition of 'consumer' needs clarification I'm not sure I feel comfortable with treating "consumer" and "producer" as relationships (as defined in the TAG Versioning note). More intuitively, IMHO, these are roles (interfaces) of (an) Agent. Defining them as "relationships" would seem to me to introduce unnecessary complexity into a definition-you need to explain what the relationship is between, which means introduction an "Act of consumption", e.g., as a separate entity (as per the Note). Strange. Kirk Wilson, Ph.D. Research Staff Member CA Labs 603 823-7146 ________________________________ From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On Behalf Of Lynn, James (Software Escalations) Sent: Wednesday, July 11, 2007 8:14 AM To: public-sml@w3.org Subject: [w3c_sml][Bug 4651] Definition of 'consumer' needs clarification Regarding the use of the terms 'consumer', 'producer', and 'understand(s)' (the comments filed with bugzilla are below) I dug around the W3 website and found the following early draft of some work the TAG is doing. [Editorial Draft] Extending and Versioning Languages: Terminology http://www.w3.org/2001/tag/doc/versioning <http://www.w3.org/2001/tag/doc/versioning> In Section 1.1 Terminology, this document briefly defines the terms 'consumer' and 'producer' in ways that I feel are aligned with the way we are currently using and intend to use the terms. The term 'understand' as relates to documents according to this and other documents in the References seems to simply mean "is able to process". It is worth noting that this document and others discuss the notion of "partial understanding" which is something we have touched lightly on in discussions, specifically that some consumers may only be able to (or choose to) process only parts of the model. There is a lot of discussion around this and I am not sure that it is all relevant to what we are discussing here, but if you want some background reading and a few pretty good analogies you can look at one of the references here: Web Architecture: Extensible languages. (See http://www.w3.org/DesignIssues/Extensible.html <http://www.w3.org/DesignIssues/Extensible.html> .) Since it seems that the TAG is attempting to standardize some of the terminology, I would suggest that we at least track the progress along these lines, i.e., consumer, producer, and understands. If this WG agrees, perhaps we should also offer our own comments to the TAG regarding the use of these terms. This seems like enough for an initial discussion within the group so I'll stop here. I would like to point out that this is marked as FPWD so if we decide to track or work with the TAG on this terminology, we may need to postpone its resolution. Regards, James Lynn W3C Service Modeling Language WG HP Software 610 277 1896 *** Comments from Bug 4651 : *** The definition of 'consumer' in section 3.1 troubles some WG members, in particular the verb "understands". In discussion during the ftf, various alternatives were proposed; what is the right way to distinguish consumers as described here from other processes or agents which might encounter an SML-IF message? Are consumers processes which understand the model? (and if so, what does "understand" mean?) Are they processes which "act on" the model (as opposed to the surface artifacts of serialization)? Are they processes which validate the model? (There seemed to be consensus that validation is NOT the right touchstone.) Further discussion of this concept is needed to reach agreement on what we want as a WG, in addition to whatever editorial effort is needed to capture that intent accurately and cleanly. In addition to the specific issue of the definition of 'consumer', the passage in question (and others) raise the related (and intertwined) question "What is it that consumers DO with SML-IF documents?" The current text says in various places that the consumer "understands" the document; this troubles enough of the WG that we should try to find some other verb that carries the right denotation and connotation and lacks the problems of "understanding".
Received on Thursday, 12 July 2007 14:32:15 UTC