- From: Wilson, Kirk D <Kirk.Wilson@ca.com>
- Date: Sat, 8 Dec 2007 07:57:07 -0500
- To: "Smith, Virginia (HP Software)" <virginia.smith@hp.com>, <public-sml@w3.org>
Ginny, The problem I see with the Full/Minimal Conformance "levels" is that those terms seem to imply endpoints of a scale (and, indeed, interoperability does seem to exist on a scale from guaranteed through "operational" to minimal or none), but the Conformance Criteria section doesn't provide any means by which we talk about degrees on the scale. (If you have "full" and "minimal" conformance, why not "partial" conformance, but what does it mean to talk about something that is *partially* conforming?) One has to read the non-normative section on interoperability to begin understanding how one might talk about the scale. Which raises another question: Does Full Conformance require embedding of all documents and schemaComplete-ness? As the definition is currently stated, I don't believe so. For example, the normative rule regarding non-embedded of documents is simply that documents that are not embedded MUST include the location element which MUST have the appropriate content. (Yes, what happened to the discussion of schemaComplete in the spec? It seems to have disappeared.) But we know that not embedding documents can adversely affect interoperability. It seems to me that we are saying conformance to the spec yields interoperability (of course) but, on the other hand, we have introduced an extension mechanism for reference schemes--and extensions always cause headaches for interoperability--and options (non-embedding of documents and schema incompleteness) that *do not violate the Schema* but adversely affect interoperability. The current definition of Full Conformance addresses only the reference scheme extension mechanism; I believe it also needs to address the other two interoperability-affecting factors that John identified if it is going to achieve the goal of guaranteed interoperability. Finally, it seems to me that having separate <referenceScheme> elements for each reference scheme that is used in a interchange set may not answer the question of whether a SML-IF document is fully conforming. Consider: <referenceScheme>sml:uri</referenceScheme> <referenceScheme>someWeirdScheme</referenceScheme> All I know is that sml:uri is used in at least one SML reference; I don't know if sml:uri is used in all references or just in a true subset of the references--and the Weird scheme is used in the others. And, in general, isn't there a problem with this <referenceScheme> idea in itself if the SML-IF producer doesn't understand a scheme: How is it to know what the <referenceScheme> should be populated with? Should the SML document using the reference scheme somehow publish the identifier (QName or URI) for the scheme? Kirk Wilson, Ph.D. Research Staff Member CA Labs 603 823-7146 -----Original Message----- From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On Behalf Of Smith, Virginia (HP Software) Sent: Friday, December 07, 2007 1:13 PM To: public-sml@w3.org Subject: RE: RE: [w3c sml] SML-IF conformance Sorry, I just realized I went too far on the conforming consumer. Use this doc instead. -- ginny -----Original Message----- From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On Behalf Of Smith, Virginia (HP Software) Sent: Friday, December 07, 2007 10:04 AM To: public-sml@w3.org Subject: RE: [w3c sml] SML-IF conformance A new draft based on feedback. Mostly some clarification on conforming consumers and the list of reference schemes. -- ginny -----Original Message----- From: Smith, Virginia (HP Software) Sent: Thursday, December 06, 2007 11:32 PM To: public-sml@w3.org Subject: [w3c sml] SML-IF conformance Attached is a draft of the SML-IF conformance proposal that we discussed today. -- ginny
Received on Saturday, 8 December 2007 12:57:21 UTC