There was an effort to do this, it was called Javascript 2, and it died in committee because the semantics would have broken the majority of existing web apps. Javascript is a late-bound language, and classes ala Actionscript add an early binding model which is compatible with the way web pages are loaded.
Yes, all improved standards submitted that I know of have died in the approval committee. That's why we need a big player to actually implement them.
And it isn't so hard, they can take one of the methods and keywords used now, and actually make them work (like prototypes), or (even better
create a few new ones on top of it that simply inplement types and inheritance.
Secondly, they do have persistent local objects. Chrome includes Google Gears, and HTML5 will add persistent local storage.
Yes, I'm waiting eagerly for one of those to become mainstream, especially GG. Although installing a pluging is a big no-no for most large companies, and so only partially useable.
Finally, if you want a scenegraph-style API, use SVG instead of Canvas. Canvas has no concept of "controls", it's a procedural, immediate mode, drawing API. It's like asking OpenGL to add Buttons, Trees, and Table widgets. It's just not relavent to the core API.
I know, but SVG isn't very usable for drawing either. That's why canvas was developed. And the only browser that natively supports SVG is IE, the rest use canvas.
The main (read: really flabbergasting) problems with webapps are, that there's no good way to store and process datasets locally (AJAX helps, but without a local store and the only container being an untyped array...), no usable way to actually draw things like your own controls (canvas helps with the drawing, but you have to move controls on top, and no realistic way to use objects with events), and the totally awful DOM model (endlessly duplicated amounts of layers and stuff, but still no good hierarchy that actually makes sense, no actually usable general purpose areas -not even DIVs- where you can DRAW your own controls on, or simply put predefined stuff on in a predictable way).
It's great that we at least have solid CSS suport across all the browsers now (read: IE), and I know about the Google canvas class for IE, but I haven't been able yet to create and draw my own stuff like with Windows apps.
And using solely CSS for positioning is a great pain when using VS.NET, the demanded development platform for most stuff.
