- From: Sebastian Schnitzenbaumer <schnitz@webaccess.mozquito.com>
- Date: Wed, 13 Nov 2002 20:15:20 +0100
- To: "Mark Birbeck" <Mark.Birbeck@x-port.net>, "'Karandikar, Shailesh'" <Shailesh.Karandikar@dendrite.com>
- Cc: <www-forms@w3.org>
Schailesh, Mark, Andrew, all, yes - good observations. There is no "problem" here - but we need experience and best practices - I will take the time and energy to expand: There is a continuum in approaching user interface, where one end of the spectrum is that the data and the UI are the same, and the other end that the two are completely different. I'm stressing the term continuum because there are many shades of grey between those two extremes - complete WYSIWYG and the UI is completely independent from the underlying data structure. Catering just for one end of the spectrum, the WYSIWYG case, would have been way to simply. For example, we could have just added the "editable" property to CSS, and assume that a XML+CSS user agent now renders everything as editable text boxes, submit the entire (changed) XML document on submission, call it XForms and be done. So we did not do that because it is an over-simplification, and even though end-users love WYSIWYG, it just too short-sighted and questions arise quickly: "how do I edit attributes?" - "where do I put my logic?". With XForms 1.0, we focussed on the harder problem of allowing developers to define their one balance between UI and data in the continuum. This was the right decision and a much more complex task to solve. What we are doing now is looking back at this edge case, or symmetry case, how I call it, in the continuum, where UI and data are in fact the same. A valid XForms 1.0 may point to itself as its instance. And how the processing is defined now, the UI would be copy/clone of the instance data. So it can be done now. We're just investigating this a little more and welcome feedback on this particular issue, however we don't have to "fix" anything. - Sebastian ----- Original Message ----- From: "Mark Birbeck" <Mark.Birbeck@x-port.net> To: "'Karandikar, Shailesh'" <Shailesh.Karandikar@dendrite.com> Cc: <www-forms@w3.org> Sent: Wednesday, November 13, 2002 6:38 PM Subject: RE: XForms CR - 3.3.2 "one and the same"? > > > One concern with making the instance point directly to the current > > document, > > rather than a seperate instance/clone of the same one is that the > > processor > > might change portions of the document for various reasons (E.g. attaching > > additional attributes to form controls, decorating instance data nodes, > > etc.) for the ease of processing. All such changes would be 'exposed' > > unintentionally. Also Some processors make structural changes to the > > document (E.g. inserting new bind elements). > > Hence the need for a publicly defined interface - my argument for a DOM > spec. You can do what you want internally, but you have to honour the public > declarations. > > Mark > > > [CARMEN] http://webaccess.mozquito.com:8080/mail/schnitz/inbox/880970921203547880.xml
Received on Wednesday, 13 November 2002 14:16:40 UTC