RE: Guide to Versioning XML Languages using XML Schema 1.1

That's great feedback.  I've been meaning to take another swag at the
doc now that fallback to declared type, auto-insertion and not-in-schema
are now in the WD, so this really helps.  BTW, auto-insertion can be
done at the schema level, and has always been possible at the "globally
when processing schema docs".

Cheers,
Dave

 

> -----Original Message-----
> From: [email protected] 
> [mailto:[email protected]] On Behalf Of [email protected]
> Sent: Friday, December 15, 2006 7:50 AM
> To: [email protected]
> Subject: Guide to Versioning XML Languages using XML Schema 1.1
> 
> 
> I took an Action to look at the "Guide to Versioning XML 
> Languages using XML Schema 1.1":
> 
>   http://www.w3.org/TR/2006/WD-xmlschema-guide2versioning-20060928/
> 
> This first Working Draft, edited by our very own David 
> Orchard, introduces a variety of techniques for evolving XML 
> messages made possible using the XML Schema 1.1 
> specification, itself a Working Draft as of 31 Aug:
> 
>    http://www.w3.org/TR/2006/WD-xmlschema11-1-20060831/
> 
> This document may be seen as a mini-Primer, introducing the 
> new features for "versioning" provided by XML Schema 1.1
> 
> Reading the document, I was unclear which examples were valid 
> XML Schema 1.0 documents, which cases relied up additional 
> features of an XML Schema 1.1 processor, and which used the 
> additional elements and attributes introduced by XML Schema 1.1.
> 
> Also in some cases I had to play "spot the difference" 
> between two similar messages, some judicious highlighting may help.
> 
> The document discusses new features of XML Schema 1.1:
> 
>   1. Weak wildcards - permits wildcards adjacent to optional elements
>   2. Updated All Group - wildcards within All Group
>   3. Negative wildcard - exclude specific namespaces and names
> 
> These operate by relaxing where xs:any may occur and adding 
> new syntax in the form of a "notQName" attribute to xs:any, e.g:
> 
>   <xs:any notQName="name:name name:given name:family" 
>       processContents="lax" minOccurs="0" maxOccurs="unbounded"/>
> 
> I'd like to see a table/guide of of new XML Schema 1.1 syntax 
> with references to the spec.
> 
> I'm unclear of the expected impact of passing the additional 
> syntax to existing XML Schema 1.0 processors. It would be 
> helpful if the document explained how to identify schemas 
> containing the new constructs if indeed the xs namespace 
> hasn't changed.
> 
> The document also cites mechanisms that are 
> possible/probable/under discussion for the next Working Draft 
> (of Schema 1.1?). I found this interesting, but less easy to grasp:
> 
>   4. Fallback To Declared Type - use declared type if 
> xsi:type is unknown
>   5. Auto-insertion of wildcards
>   6. Not in Schema wildcard - a wildcard that allows anything 
> not defined in the current schema
> 
> Of these, personally I think (5) and (6) are particularly 
> interesting as they go towards final proposals and 
> discussions made under our LC124 which changed how an 
> existing XML Schema 1.0 document may be processed differently 
> to allow more open content.
> 
> Except, AIUI, the examples all use a new "openContenct" (name 
> not fixed) model group, e.g.:
> 
>   <xs:openContent namespace="##any" processContents="lax">
>      <xs:element .../>
>      <xs:element .../>
>   </xs:openContent>
> 
> I didn't spot a way to engage this flexibility at the schema 
> level, or globally when processing existing Schema documents, 
> which must be how many people would like to work. If this 
> isn't allowed, hightlighting the restriction would be useful.
> 
> I personally find the use of xsi:type (4) rather worrying:
> 
>    http://blog.whatfettle.com/2006/11/29/xsitype-is-evil/
> 
> There are also placeholder sections for:
> 
>   12.5 Version numbers
>   12.6 MustUnderstand
> 
> Overall I'm concerned at the wide variety of new features and 
> techniques being introduced, and the strong likely hood how 
> few will be implemented in a consistent manor by 
> implementations across the board, but it's good to see them 
> enumerated in a single focussed document.
> 
> The document lacks a reference section, cites the 31 August 
> XML Schema 1.1 WD, but doesn't AFAICT link to it.
> 
> --------------------------------------------
> Proposed WG Actions:
> 
> We should first congratulate the XML Schema WG for pursuing 
> this very important topic, of interest to the WSD WG and 
> providing what appears to be a number features directly 
> targeted at enabling the description of Web services messages 
> which may evolve.  
> 
> We should reference this document as from the Primer in the 
> list of works "relevant to versioning of Web services description"
> in Section 5.2.
> 
> We can also now remove the Primer ednote: 
> 
> """
> 20050706 This section may be subject to change dependent upon 
> the outcome of the WSDL Last Call Issue LC124, which 
> discusses support compatible evolution of messages described 
> using XML Schema 1.0.
> """
> 
> Let's plan to review the note formally when it's closer to 
> final publication.
> 
> Paul
> 
> 
>