yes now I understand your point. well I think the author of j2s-swt had to fill somehow the breach between the "dektop visual paradigm" and web's. IMHO in this case a desktop simulation was made for that. Another would be to modify the concept of Shell (swt root widget) adapting it to a html document. personally, I like the current approach.
Another important difference between desktop and web, is the position of components. In all desktop gui frameworks I know there is *a single* positioning mode (absolute) and layout objects are responsible of setting a widget children's bounds. On the other hand, in the "web visual paradigm" css define several positioning modes (in my opinion the more difficult part of the standar to understand and implement) and there are not layout objects (in the web, the layout is defined in the standar while in desktop the layout is defined by an object).
In j2s-swt, all html elements representing swt widgets use css position:absolute (I think this is the only way "desktop visual paradigm" can be "simulated" in an html document) and this makes very difficult for "embedding" a swt component inside an existing html document (for example your blog) that normally contains inline/block/table positioned elements.
IMHO, one of the desired characteristics of j2s (the main attraction) is the possibility of using visual desktop ways/paradigms/components for developing guis (for example, using a gui visual editor). and I don't think j2s must sacrifice this characteristic in benefit of adaptability/iteraction with web applications. But I'm agree with you that perhaps some desktop concepts (mainly window and window events) can be adapted in some degree to "live" inside html docs.
--
Sebastian Gurin <sgu...@softpoint.org>