- From: Wilson, Kirk D <Kirk.Wilson@ca.com>
- Date: Sat, 15 Sep 2007 15:23:11 -0400
- To: "John Arwe" <johnarwe@us.ibm.com>, <public-sml@w3.org>
- Message-ID: <F9576E62032243419E097FED5F0E75F303276209@USILMS12.ca.com>
After much mind-searching, I believe the point was that an instance document might be subject to inconsistent constraints such that if it satisfies one of the constraints, it would violate the other and vice-versa. You could never have an instance document that satisfies both constraints.
In the absence of a formal proof of rule consistency-which I'm assuming does not exist-the inconsistence would only show up only if you have an instance document that causes the incompatible constraints to be invoked.
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 John Arwe
Sent: Thursday, September 13, 2007 8:36 AM
To: public-sml@w3.org
Subject: RE: [Bug 4638] Conformance section needed - consistency of rules
What Kirk attributed to me _really_ is not ringing any bells and the language doesn't sound right for me. If someone can cite a section from the minutes maybe we can figure out what (maybe who) he is referring to. If this was the June F2F, at about that time I was prototyping a use of SML for IT modeling where we quickly concluded that reverse-parsing Schematron expressions to figure out "intent"... which is required for the consistency test under discussion as I am interpeting it... was foolishness incarnate. So I would have had that bruise fresh on my mind and would have steered people away from it, not toward it.
If this was the August F2F, I plead Roberto Gonzales.
Again, if someone can point to a snippet of minutes bearing on this discussion that might help. It's always possible this page is out on drum somewhere under an alias. If not, we might choose to apply my "if it's important, it will come back" rule.
Best Regards, John
Street address: 2455 South Road, Poughkeepsie, NY USA 12601
Voice: 1+845-435-9470 Fax: 1+845-432-9787
Kumar Pandit <kumarp@windows.microsoft.com>
09/13/2007 02:05 AM
To
Pratul Dublish <Pratul.Dublish@microsoft.com>, "Wilson, Kirk D" <Kirk.Wilson@ca.com>, "Smith, Virginia (HP Software)" <virginia.smith@hp.com>, "public-sml@w3.org" <public-sml@w3.org>, John Arwe/Poughkeepsie/IBM@IBMUS
cc
Subject
RE: [Bug 4638] Conformance section needed
I agree. I don't think we can perform static analysis to determine if two rules are conflicting in every case. May be John meant something else when he referred to consistency. John, can you please shed some light on this?
From: Pratul Dublish
Sent: Wednesday, September 12, 2007 10:26 PM
To: Wilson, Kirk D; Kumar Pandit; Smith, Virginia (HP Software); public-sml@w3.org
Subject: RE: [Bug 4638] Conformance section needed
The Schematron spec does not cover consistency. Schematron is query language agnostic, and so Schematron constraints can be written in query languages that are Turing complete (i.e. can describe the behavior of a Turing machine). So determining that two Schematron rules are inconsistent is likely to be an undecidable problem. Even the default query binding (extended version of XPath 1.0 as defined in XSLT) allows constraints with extension functions and hence determining that two rules are inconsistent is likely to be undecidable.
From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On Behalf Of Wilson, Kirk D
Sent: Wednesday, September 12, 2007 4:30 AM
To: Kumar Pandit; Smith, Virginia (HP Software); public-sml@w3.org
Subject: RE: [Bug 4638] Conformance section needed
I believe it was John who, during the F2F, mentioned a critical issue, namely, the consistency of rules in the model. A model is invalid if it has inconsistent rules. Consistency seems to me to be an important point to cover in the conformance criteria. (Is it covered by ISO/IEC 19757-3?) It seems to me that formally testing the consistency of a set of Schematron rules is a substantial theoretical challenge. Are their known algorithms for doing that?
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 Kumar Pandit
Sent: Saturday, September 08, 2007 1:16 AM
To: Smith, Virginia (HP Software); public-sml@w3.org
Subject: RE: [Bug 4638] Conformance section needed
Ginny, thanks for making the relevant changes. I agree that some of the changes can wait until after the second draft.
From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On Behalf Of Smith, Virginia (HP Software)
Sent: Friday, September 07, 2007 4:04 PM
To: public-sml@w3.org
Subject: RE: [Bug 4638] Conformance section needed
I think that we do need to check the content against the rest of the spec and revise as necessary - I thought maybe after Second Draft would be a good time - but it does not seem like the time is right just yet. For the time being I mostly just reworded and reorganized the section. We could be more vague and, instead of mentioning each requirement, just refer to the normative sections (which are not clearly identified as such yet).
I agree that a bulleted list is appropriate so the conformance criteria can be identified. I will change that.
The terminology in section 2.2 doesn't exactly match. I think we should make the change suggested by Kumar (the 2nd #2) but I would like to also change the terminology to say "instance document" and "definition document".
--
ginny
________________________________
From: Kumar Pandit [mailto:kumarp@windows.microsoft.com]
Sent: Friday, September 07, 2007 12:07 AM
To: Smith, Virginia (HP Software); public-sml@w3.org
Cc: Kumar Pandit
Subject: RE: [Bug 4638] Conformance section needed
Ginny's proposed text looks good and I agree with the suggestions from Kirk and Sandy.
Do we need additional items listed below? I am not proposing the actual text here.
1. Each SML reference in each model instance document must satisfy reference validity conditions as defined in [add reference to section where we define 'at most one target' etc. conditions].
2. Each instance document in the model MUST satisfy all applicable SML identity constraints.
3. If SML identity constraints are defined in a schema document, they must be defined in conformance to section 4.4 Identity constraints (for example, verify that they appear inside xs:appinfo section and that the selector/field xpath matches the BNF we define, etc.)
4. Similarly for schematron constraints.
Other suggestions:
1. The list should be a numbered list instead of a bulleted list.
2. Items should specifically say 'each instance document' or 'each schema document' instead of 'each document'.
From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On Behalf Of Smith, Virginia (HP Software)
Sent: Thursday, September 06, 2007 10:53 PM
To: public-sml@w3.org
Subject: RE: [Bug 4638] Conformance section needed
I agree with Sandy's suggestion. I'll make the change.
________________________________
From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On Behalf Of Pratul Dublish
Sent: Thursday, September 06, 2007 8:29 AM
To: Sandy Gao; public-sml@w3.org
Subject: RE: [Bug 4638] Conformance section needed
Agree with Sandy and Kirk
From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On Behalf Of Sandy Gao
Sent: Thursday, September 06, 2007 7:03 AM
To: public-sml@w3.org
Subject: RE: [Bug 4638] Conformance section needed
+1 to Kirk's comment.
Maybe it'll be even better to treat acyclic in the same way as target*. e.g.
* Each document in the model MUST satisfy all applicable Schematron constraints
* Each document in the model MUST satisfy all applicable sml:acyclic and sml:target* constraints
We don't have to repeat "no cycle" here. That rule is already specified elsewhere in the spec.
Thanks,
Sandy Gao
XML Technologies, IBM Canada
Editor, W3C XML Schema WG <http://www.w3.org/XML/Schema/>
Member, W3C SML WG <http://www.w3.org/XML/SML/>
(1-905) 413-3255 T/L 969-3255
"Wilson, Kirk D" <Kirk.Wilson@ca.com>
Sent by: public-sml-request@w3.org
2007-09-06 06:27 AM
To
<bugzilla@wiggum.w3.org>, <public-sml@w3.org>
cc
Subject
RE: [Bug 4638] Conformance section needed
Now that we have dropped sml:refType, is "reference type" (see last bullet, last phrase) a well-defined term in the spec? I would think we need to give an explicit definition of what we intend to cover with this term or, simpler, just drop the word "type" from the phrase.
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 bugzilla@wiggum.w3.org
Sent: Thursday, September 06, 2007 5:56 AM
To: public-sml@w3.org
Subject: [Bug 4638] Conformance section needed
http://www.w3.org/Bugs/Public/show_bug.cgi?id=4638
virginia.smith@hp.com changed:
What |Removed |Added
----------------------------------------------------------------------------
AssignedTo|cmsmcq@w3.org |virginia.smith@hp.com
Keywords|editorial |needsReview
------- Comment #3 from virginia.smith@hp.com 2007-09-06 09:55 -------
In the SML spec, changed the "Model Validation" section to "Conformance
Criteria" as follows:
A program is a conforming SML model validator if it satisfies the following
conditions:
* The validator MUST perform model validation as defined in this
specification. Model validation is the process of examining each document in a
model and verifying that this document is valid with respect to the model
definition documents, i.e., each model instance document satisfies the schemas
and rules defined in the applicable model definition documents.
* The validator MUST support XML Schema 1.0 and XPath 1.0 but MAY also
support any future versions of these specifications.
* The validator MUST perform Schematron rule evaluation on the #ALL
phase, implying that every rule in every pattern is evaluated.
A set of XML documents is a conforming SML model if it satisfies the following
conditions:
* Each document in the model MUST be a well-formed XML document [XML
1.0]
* Each XML Schema document in the model's definition documents MUST
satisfy the conditions expressed in Errors in Schema Construction and Structure
(§5.1). [XML Schema Structures]
* Each Schematron document in the model's definition documents MUST be
a valid Schematron document [ISO/IEC 19757-3]
* In each instance document in the model, the [validity] property of
the root element and all of its attributes and descendants MUST NOT be
"invalid" when schema validity is assessed by a conforming schema-aware
processor with respect to the referenced XML Schema documents in the model's
definition documents. [XML Schema Structures]
* Each document in the model MUST satisfy all applicable sml:target*
and Schematron constraints
* The model MUST NOT contain a cycle whose edges are references of
type R if R is an acyclic reference type
Received on Saturday, 15 September 2007 19:23:18 UTC