Taking a closer look at what is XAML, and jetlag leaving me enough time to think about it, I don't find anything really new here for Microsoft and think only festa-like journalists can believe there is strong innovation here. Unordered thoughts:

  • MSIE has a long history of HTML-based dialogs. At the beginning of 1997 during the WWW conference in Santa Clara, Jean Paoli and Adam Bosworth confirmed me the dialogs of IE were coded in HTML. So that means at least since IE4. It seems to me that XAML elements are more or less classes of a mix of HTML form and HTML container elements. HTML forms on windows use the native widgetry anyway, like XAML. No need to reinvent the wheel here.
  • The fact that XAML elements are styled by attributes and not by CSS is another possible indicator that what we really have here behind XAML is HTML. From a MSFT point of view, it's kind of common sense styling the elements through attributes and not CSS : XAML rendering is then defined by one spec only, fully in the hands of MSFT and does not rely on CSS, a spec edited by a non-MSFT entity called - berk - a Consortium where various interests including challengers - kill'em kill'em - can impact a spec. Anyway, that's is definitely not a modern way of thinking and this announces more trouble for the W3C in our world of standards, unless Microsoft explains in a developer doc that there is a bijection between XAML attribute and their CSS counterparts.
  • Corollary, this is a big stone in the garden of XForms. XForms suck, really suck. Despite of a great power and existing/running implementations, it's incredibly complex and I wish good luck to Web authors who want to make intensive use of XForms. Solutions like XAML and XUL, even if less powerful, show an adaptability and a maniability that XForms is far away from. Microsoft is clearly showing where it stands here. And to my own suprise, I think it's good. But there is nothing new here for MSFT.
  • we first learnt about the implementation of a behavior mechanism called HTC (including binary behaviors) around the 10th of June 1998.... More than five years ago... So compiled code attached to a markup defining the behavior of an element is ooooold news in the Microsoft world. It really seems that compiled code attached to XAML files are nothing more nothing less than a recooking of binary HTC with a .NET sauce. From my personal perspective, that does not look fresh meat at all but I could be wrong here since I find Visual Studio a quite good example of almost correct software quality at MSFT. The people in the languages/compilers division seem to be quite innovative and productive.

In conclusion, I think that XAML = HTML + one HTC to bind XAML elements and attributes to HTML and CSS + binary HTCs for the behaviors.

A superb HTML+JS color picker

Neil Marshall has made a superb colorpicker in HTML + JS. He proposed me to have a XUL version of it for Composer and I think it is a great idea.

By the way....

BTW, 01-nov-2003 was officially my last day at AOL even if left the office mid-August. I am no longer an employee and gained back my freedom of speech :-)