I think there's a tension right now in the web platform community about whether we're in the post-HTML+js world yet. Cappuccino, GWT, jqueryui, etc are all moving in that direction. But there's a tension between libraries that want to treat HTML+js like dumb rendering layer, and libraries that keep HTML+js as the developer-level workspace, and just extend it by adding API.
There are a lot of things that work really well in HTML+js (links, urls, scrolling, etc) that get screwed up as you add more post-HTML widgetry. So the developers of post-HTML tools eventually end up rewriting everything HTML did in their new toolkit. The toolkits get better and better, but I have yet to see one that didn't break something that HTML did well.
So it becomes an either/or proposition. Either I jump fully into Cappuccino, accept that it will break things about HTML and I will not be able to fix them, or I stay in the land of HTML, and I try to implement the cool stuff Cappuccino does (windowing, OO, advanced widgetry, etc) with HTML+js and whatever JS libraries I can find, but generally sick with The HTML Way wherever possible.
The thing is, I know what HTML sucks at. I have no way to know what Cappuccino breaks that HTML does well until I am waist-deep in a Cappuccino app. I think that makes it hard for me to jump.
And so my preference (although it's just a preference) is for small tools that add one thing to HTML+js, but they do it really well. So if I want to make a mobile app for iPhones, I want to start with a JS library that lets me make touch-slideable panes, that's incredibly fast and smooth, and that's it. I don't need it to create a whole complex windowing model, or re-implement HTML buttons in javascript, or anything like that.
But I feel like the tide is slowly turning... the post-HTML frameworks are getting better. At some point one will be good enough that I really can leave HTML behind. Maybe I'll try out Cappucino for my next weekend project and see how it goes.