I've been mulling over an architecture to html5 enable Premise. Much of it comes from my learning on my HB UI project. Since there is a chance one day I will move off HB and maybe onto Premise, this is somewhat of a x-post to my thoughts on a HB html5 UI.
On the client side...
Every classt in Premise will have a HTML5 widget. There are plenty of html design tools out there that allow you to use your own widgets. The widget will be derived from a javascript framework such as jquery or dojo or ???. The widgets will also be required to work with other components of the selected framework. This will enable all sorts of goodies on the selected framework such as websockets, subscribe/publish changes, push technology, cross platform, + + +.
On the server side...
A websocket server will need to be built for delivering the data, the current server will probably work for delivering the html5 and css. There seem to be plenty of .net or C websocket libraries out there could help out in this area.
The key to it all is Premise widgets. If a widget is created for every class, even new classes that people create, it makes for a nice pluggable experience on the UI. Of course module creators will have to write or adapt their own widget for any new classes they create, but as long as the Premise widget is compatible with dojo, yui, jquery, openajax, or whatever framework is selected, it should work with the underlying framework that would tie it into Premise. I suspect most folks would be 99% covered with "stock widgets" that would corrspond to stock classes in Premise.
It's a "cocktail napkin" idea right now, but I think all the components are there... HTML5, design/layout, websocket push, subcribe/publish, pluggable, extensable.
Thoughts?
Tim