Re: [Moderator Action] Schema Modularisation in XForms (Was: xslt stylesheet for xforms to xhtml)

Moderator: sent to www-forms@w3.org and www-forms-editor@w3.org 
  ----- Original Message ----- 
  From: AndrewWatt2001@aol.com 
  To: MSeaborne@origoservices.com ; xforms@yahoogroups.com ; www-forms@w3.org ; www-forms-editor@w3.org 
  Sent: Monday, September 02, 2002 3:58 PM
  Subject: [Moderator Action] Schema Modularisation in XForms (Was: xslt stylesheet for xforms to xhtml)


  In a message dated 02/09/2002 10:31:19 GMT Daylight Time, MSeaborne@origoservices.com writes:



    That is interesting. The thing that attracts me to XForms is not really the UI bit, but the model. I think this is where the key prospect for reusability lies. I work for a message standards body, and we currently maintain our own XForms like language, that companies in the industry use to build forms based applications. It makes a half-hearted attempt to separate the UI from the model, but fails. So our forms have to be authored entirely with a specific renderer in mind. My hope for XForms is that the model should become fully portable across implementations.

    My main niggle with XForms is that the binding/validation functionality seems to be there largely to compensate for lack of same in W3C XML Schema; which makes that bit of XForms into a schema language. We have ourselves recognised problems with the expressiveness of W3C XML Schema, and at the moment all the schemas we publish are supplemented by message implementation guidelines; prose documents which tell you about all the data dependencies and validation rules about which W3C XML Schema remain silent. Rather than inventing our own vocabulary to add to what W3C XML Schema already does, we are instead investigating the use of existing schema languages to achieve this. The most obvious candidates to fulfil this function are Relax NG and Schematron, especially as these have, or are about to receive official status from ISO. I do not know for certain, but I strongly suspect that the majority of what is currently prose will in the future be expressed in one or other of these sc! he! ma languages.

    The data dependencies and validation rules that an XForms application must be capable of enforcing are, at least for the industry in which I work, exactly those which an application generating XML instances without human intervention must enforce (if I am over-simplifying, I would be really grateful if someone would put me straight on this). The differences arise only at the implementation level, it seems. So my hope is that one bundle of schemas will form the basis of any application required to process/generate our messages. 

    Which begs the question: should XForms be adding yet another language for expressing schema-like constraints on XML documents? Given that XForms is required to function without external schemas at all, then the obvious answer is yes. Naturally, I would be very happy if XForms, in addition to what it already does, provided me with a method for linking instances to a) more than one schema; b) expressed in more than one schema language. However, I am reconciled to the fact that, at some point in the not too distant future, I will be looking to provide mappings between the bits that XForms adds to validation and any equivalent functionality provided by non-W3C schema languages.

    All the best

    Mark Seaborne

    Origo Services Ltd


  Mark,

  You raise several interesting points relating to schema support in XForms.

  It seems to me that it would be very useful to have modularisation of schema (the generic term) support in XForms, rather than a "hard-wired" need for W3C XML Schema support.

  For example, allowing Relax NG to be used, might open up XForms to a range of mobile browsers at the limited functionality end which would have difficulty in supporting XForms because it has difficulty in implementing W3C XML Schema.

  At the opposite end of the spectrum, it might be useful to be able to plug in Schematron to apply rule-based business logic that W3C XML Schema, at least in version 1.0, can't handle.

  If XForms is to be universally accepted as a forms language which is generally useful then it seems to me that it should allow a developer to choose a schema language which meets a particular set of business needs, whether or not those happen to be expressible in W3C XML Schema. I guess W3C XML Schema, as applied in XForms, could be interpreted as locking a developer in to a proprietary schema language.

  Quite possibly these are questions we should have raised a long time ago but it nonetheless seems to me to be good design to allow components - including schema validation components - to be snapped in or out of an XForms design.

  Andrew Watt