RE: [dita] W3C XML Schema 1.1

dita message

Thanks for the prompt response. In the DITA Open Toolkit 1.0.1 readme we found the following statement regarding the DITA XSD implementation:

This document describes the release notes and usage of the experimental DITA XML Schema on developerWorks®.

This DITA release includes an implementation of the topic architecture in XML Schema. The DTDs are still the canonical representation of DITA. The design pattern for the DITA Schemas is based on the W3C XML Schema 1.0 Specification and may be subject to change in the future.

The specialization process and design pattern for the DITA XML Schema are still being developed and refined. As such, the specialization process stated below should not interpreted as "the definitive process", but simply one method to specialize information types and domains

However, what I understand from your response, DITA does not depend on the inheritance model of XSD today, and may not in the future either. As such, our implementation of DITA XSD can go ahead without relying on future evolution of the XSD standardization process.

Nevertheless, if future release of DITA does incorporate future XSD enhancements on the inheritance model, we will need to do migration if we want to adopt such future releases. Could you please give us a sense how much a migration effort will that be?

-----Original Message-----
From: Eric Sirois [mailto:esirois@ca.ibm.com]
Sent: Friday, April 29, 2005 8:07 AM
To: Tsao, Scott
Cc: dita@lists.oasis-open.org
Subject: RE: [dita] W3C XML Schema 1.1

Hi Scott,

The current incarnation of the DITA Schemas fully adheres to the XSD 1.0 spec and has no dependency of the XSD 1.1. The schemas were redesigned from top to bottom in the developerWorks 1.3.2 to remove any architectural dependency that would only appear in XSD 1.1. Otherwise, most commercial/open-source XML tools would not be able to validate instance documents using the DITA Schemas.

If DITA needs requirements for XSD 1.1 (RQ-11) , that would mean that the DITA Schema would make use of the XSD inheritance mechanism. Which also means that the DITA would have to be redesigned from top to bottom. That does not necessarily mean that DITA will use that feature.

RQ-11 is only an issue *if and only if* we make use the XSD inheritance model. This is a very important point. If DITA is implemented using the inheritance model XSD still would not be able to fully implement DITA. What's missing in the XSD inheritance architecture to fully support the DITA architecture? Renaming or aliasing element in a specific content model (local element substitution). XSD only supports global element substitution via substitutionGroups.

There is a possibly a way that we may be able to achieve local element substitution via the XSD inheritance model, but implementing it in XSD architecture would be very ugly!

We can model the DITA 1.0 and 1.1 using XSD 1.0. We currently have no dependency on any XSD 1.1 features. We may have a dependency on XSD 1.1 features for DITA 2.0, but it will probably not involve any feature that is dependent on the XSD inheritance architecture.

I hope this helps address your concerns.

Kind regards,
Eric

I've included below a response to question regarding *issues* with the DITA Schemas on the dita-users list (just in case some folks on the TC may not be subscribe to that list).

<snip>

There are no issues, per se, with the DITA Schemas. The schemas are fully equivalent to the DTDs and work fine with most XML authoring editors...IDE's are another issue. :-) . The *issue* with the W3C Schemas is that we are able to take full advantage of the inheritance capabilities. Why do we want to use the W3C XML Schema inheritance capabilities? Well because their inheritance rules are pretty close to the DITA inheritance rules. And if were to make use of that it would be easier for developers to create valid specializations that conform to the DITA spec.

We developed schemas over a year ago that *tried* Schema inheritance, but kept hitting wall after wall of issue in the manner some parsers adhered to the XML Schema 1.0. Some were very strict while other were lax. We could not get a consistent story. Since the developerWorks DITA 1.3.2 version and the spec version that is available at OASIS we have removed all use of inheritance in the schema, but also need to come up with an architecture that was *fairly easy* to use so developers/authors/architects could create new specializations. Unfortunately, they still (along with the DTDs...and same would occur with RNG) manually verify that the specialized element's content is valid as per the DITA spec.

Probably one of the biggest reason that people are still using the DTDs is that XSLT 1.0 processors only support DTDs out-of-the-box. Web browsers only supports DTD parsing. If you look at a schema validate instance in a browser and trying to use a stylesheet to view the doc...you not going to see what you were expecting. If you want to see a schema validated instance doc in a browser you have to use CSS. There are no free XSLT 2.0 (Schema aware) processor available at this time. You have to pay for Saxonica-SA and the XMLSpy one. I believe Stylus Studio use Saxonica.

The entire DITA processing system is dependent on a default class attribute value that appears on every element in DITA, like <xs:attribute ref="class" default="- topic/topic "/>. All instance docs need to validated so that the class attributes are normalized...then and only then can the instance document be processed by the XSL stylesheets.

If your asking yourself...how do I process my instance documents if they are all schema validated? Well..in the demo directory in the DITA open source project there is a java directory...in it you will find a couple of small applications, TransformUsingXMLSchema and ValidateXMLDoc, that can help those using XML Schemas or want to use XML Schemas to process their documents.

TransformUsingXMLSchema does what the applications says..it can transform XML Schema validate docs using XSLT 1.0 processors. All the application does is validate and normalize the class attribute and then pass the doc to XSLT processor to be processed.

ValidateXMLDoc allows you to validate an XML instance doc with an external schema when the instance has a DOCTYPE defined in it. You don't need to modify your multitude of legacy docs that have the DOCTYPE in it for them to be validated. Unfortunately, this option is implemented in TransformUsingXMLSchema yet. If someone has a requirement or would like to see the feature added to it...let us know or open a feature request on SourceForge.

I hope this helps clear up some of the *issues* regarding XML Schemas...or create new ones. ;-) </snip> Eric A. Sirois Staff Software Developer DB2 Universal Database - DBA XML Tools Development IBM Canada Ltd. - Toronto Software Lab
Email: esirois@ca.ibm.com
Phone:(905) 413-2841
Blue Pages (Internal)

"Transparency and accessibility requirements dictate that public information and government transactions avoid depending on technologies that imply or impose a specific product or platform on businesses or citizens" - EU on XML-based office document formats.


             "Tsao, Scott"
             <scott.tsao@boein
             g.com>                                                     To
                                       Eric Sirois/Toronto/IBM@IBMCA
             04/28/2005 06:35                                           cc
             PM                        <dita@lists.oasis-open.org>
                                                                   Subject
                                       RE: [dita] W3C XML Schema 1.1







Eric,

Thanks for the reply. I noticed that this feature has been included in the XML Schema 1.1 requirements (see http://www.w3.org/TR/xmlschema-11-req/#N40014F). However, since we are using the W3C XML Schema 1.0 spec (not DTD) for most (if not all) of our XML projects at Boeing, I am concerned if we are taking a major risk to adopt DITA today before the following events actually happen:
1) W3C releases the XML Schema 1.1 Recommendation
2) DITA incorporates the needed feature(s) of XML Schema 1.1

Please help us better understand the potential risks.

Thanks,

Scott

-----Original Message-----
From: Eric Sirois [mailto:esirois@ca.ibm.com]
Sent: Monday, September 20, 2004 11:10 AM
To: Tsao, Scott
Cc: dita@lists.oasis-open.org
Subject: Re: [dita] W3C XML Schema 1.1

Hi Scott,

The issue related to "restriction via subsumption" is still active within the XML Schema WG.

RQ-17 - http://www.w3.org/XML/Group/2002/07/xmlschema-1.1-current-reqs-list.html
#restrictn-rules
(not sure if you have to be a W3C member to view the document).

I did speak to IBM's rep on the workgroup and he's is aware of our requirements as well as the WG. They working on solving this issue for WXS 1.1. Part of the new algorithm is implemented in Saxon 8.0 and XSV

Kind regards,
Eric
Eric A. Sirois
Staff Software Developer
DB2 Universal Database - DBA XML Tools Development
IBM Canada Ltd. - Toronto Software Lab
Email: esirois@ca.ibm.com
Phone:(905) 413-2841
Blue Pages (Internal)

             "Tsao, Scott"

             <scott.tsao@boein

             g.com>
To
                                       <dita@lists.oasis-open.org>

             09/15/2004 02:09
cc
             PM

Subject
                                       [dita] W3C XML Schema 1.1

I recall that someone has mentioned that DITA requires functionality beyond the current XML Schema (1.0) spec, and I wonder if those requirements have been documented somewhere and are being considered by the W3C WG to be included in the XML Schema 1.1 spec.

Regards,

Scott Tsao
The Boeing Company
P.O. Box 3707, MC 67-EF
Seattle, WA 98124-2207
425-234-3761
scott.tsao@boeing.com