So GWT best practices encourages one to use some flavour of MVP, which should in theory allow one to write different native views while sharing the presenter business logic.
This seems to be at the heart of the GWT spin off Google project http://code.google.com/p/j2objc/ which converts the non-UI part of your code to Objective-C, allowing you to write the rest natively in Objective-C.
So my question is: If this really hard part of the puzzle is being solved, how hard would it be to include an HTML5 mobile library (like MGWT or Touch4j [Sencha]) into this MVP pipeline to have the best of all worlds?
Having dabbled with http://code.google.com/p/playn/ , this clearly seems to be the blue-print for having a cross-plaftform build system (native android & html5 & java &...), but that project is geared for single screen drawing and event loop for game dynamics and doesn't allow for keyboard input and other typical mobile goodies.
It seems a shame that if so much of the problem has been solved, that it's not possible to go the extra mile. The answer to this question would be the best plan for actioning a solution, including such nigglies as which MVP structure to choose that would ease accommodation of the various widget libraries (GWTP vs MVP 2.1), and if the best approach is to start with the PlayN code base, and start to hack it.. what are the gotchas? Or if another path is chosen, why that one? and why would it be the best??
Thanx a lot. :-)