1. 2005 W3C XML Schema Workshop Path Forward Soumitra Sengupta Microsoft

2. Office Customers Scenarios Documents are integrated in Business Processes – How do we represent information? Documents or Databases? Tax, Receipts, Birth Certificates, Driver License, Contract, Proposal, Insurance, Mortgage, Budget Proposal, Status Report, Meeting Minutes, Functional Specification, Customer Inquiry, Project Bid, Project Schedule, Bill of Materials,, Customer Pitch, Project Timeline Diagrams, Org Chart Diagrams, Furniture Selection Book – Same Schema language in order to Represent to End User the “Insurance Proposal Document” Integrate the Insurance Proposal in a Back-End (in WSDL, in Database) Co-occurrence constraints is of interest to Office

3. Office Experience - WordWord Supports full XSD spec Recommendations – Define extensible namespaces - Encourages use of Internet URL, followed by an internally-agreed-upon naming convention. – Use a single element component to define the root element, and use a series of complexType components to define all of the root element's descendant elements – Embed documentation directly into the schema – Use the minOccurs and maxOccurs attributes wherever possible to strictly define the number of times that a component occurs – Use strict data typing wherever possible – Restrict the range of input values wherever possible

4. Office Experience - ExcelExcel Infers schemas when no schemas provided – Look at document for the rules for inference and caveats Not supported (for now in Office 2003) – element – anyAttribute – Substitution group – Mixed content – Abstract element – Recursive structures more than one level deep Recommendations – strong data types – has full attribute & element

5. Office Experience - InfoPathInfoPath Not Supported – DTD’s – Abstract ComplexType Reduced Support (we found for them good UI paradigms) – Required xsd:any (w/minOccurs > 0) At design time gives you a choice about element – Substitution Groups – now supported with SP1 – Unbounded Choice – now supported with SP1 – Repeating Sequence – now supported with some limitations in SP1 – Choice of Model Groups – now supported with some limitations in SP1 Special Meaning – Adding New Element Fields and Groups with the Data Source Task Pane – Adding New Attribute Fields with the Data Source Task Pane – Storing XML Signatures in the Data Source – Binding a Field to a Rich Text Box Control XPath comparison (similar to co-occurrences) in InfoPath

6. Links to the articles And in general a lot of XML dev articles has been posted on – http://msdn.microsoft.com/office/understanding/xmloffice/articles/ http://msdn.microsoft.com/office/understanding/xmloffice/articles/ Experience using XSD with Word – http://msdn.microsoft.com/office/understanding/xmloffice/articles/default.aspx?pull=/library/ en-us/odc_wd2003_ta/html/odc_wdalpine.asp http://msdn.microsoft.com/office/understanding/xmloffice/articles/default.aspx?pull=/library/ en-us/odc_wd2003_ta/html/odc_wdalpine.asp Experience using XSD with Excel – http://msdn.microsoft.com/office/understanding/xmloffice/articles/default.aspx?pull=/library/ en-us/odc_xl2003_ta/html/officeexcelxmlmappingscenarios.asp http://msdn.microsoft.com/office/understanding/xmloffice/articles/default.aspx?pull=/library/ en-us/odc_xl2003_ta/html/officeexcelxmlmappingscenarios.asp – http://msdn.microsoft.com/office/understanding/xmloffice/articles/default.aspx?pull=/library/ en-us/odc_xl2003_ta/html/officeexcelinferdetails.asp http://msdn.microsoft.com/office/understanding/xmloffice/articles/default.aspx?pull=/library/ en-us/odc_xl2003_ta/html/officeexcelinferdetails.asp Experience using XSD with InfoPath – http://msdn.microsoft.com/office/understanding/xmloffice/articles/default.aspx?pull=/library/ en-us/odc_ip2003_ta/html/odc_infinfopathxsd.asp http://msdn.microsoft.com/office/understanding/xmloffice/articles/default.aspx?pull=/library/ en-us/odc_ip2003_ta/html/odc_infinfopathxsd.asp

7. We feel the pain When Doug cannot handle the complexity, guess who he passes it to It has taken us 5 years to get decent comprehension We have invested heavily on schema compatibility work in System.Xml and MSXML – Both products will be shipping later this year – We will support these products (and others that use them) for the next 10 years After years of work, we have a comprehensive test suite We have one or more product teams dealing with most of the problems we have discussed

8. Key pain points Complexity of spec Writing style just compounds complexity Incompatible implementations Versioning UPA X O and X R mapping

9. We have paid a heavy price BEFORE AFTER

10. There is progress as well XSD is ubiquitous and has strong momentum – A huge set of schemas created by our customers (300 thousands are developing XML Office 2003 solutions, 1 million developers for Office overall) – On last check we know about 22,000 schemas (*not counting* customers of Office 2003) – XBRL (uses substgrp) getting traction Number of products and protocols supporting XSD / profile of XSD is growing Office (Word, Excel, Powerpoint, InfoPath) adopting XML 1.0 and XSD 1.0 is increasing adoption further Oracle, DB2 and SQL Server support XSD 1.0 is providing real value in real customer scenarios Disruption now is the last thing we need

11. We can and should do more But XSD 1.0 Recommendation should continue to define the basic foundation for interoperability – Vendors should work hard to implement spec correctly and completely – Focus on clarifying ambiguities and correcting actual bugs via the errata process – Better documentation – Show the courage to oppose bringing any breaking changes in the XSD 1.1 proposal forward to Recommendation status at this time Problems with data binding should remain out of scope for W3C. Let vendors evolve languages to deal with this issue better Test Suites Profiles for broad scenarios (eg WS-I) Publish best (not bad) practices gained from actual deployments

12. “perfection is enemy of good” soumitrs@microsoft.com